DiAVC decoder, yet another fast H.264 high profile decoder

[schweinsz]

stage0
stage1
DiAVC init successed
start receive
start receive
start receive ... ("start receive" loops forever)

no crash
Ok, but do you observed the output of frames? Do the MPC-HC keep black screen forever?

[schweinsz]

For me it didn´t crash (on Windows XP x64, Athlon 64 X2). I also checked the output file, seems ok, only the frame order doesn´t seem to be ok on the file you provided. I also tested some other files, no problem with them. The new version makes no difference for me, only more verbose output.
If the win32.exe works fine, I believe the frame order is OK. The sequence is one of my test sequences, I often use it to verify the correctness of my decoder.

May I ask when you fix the comptability problem of DiAVC with DVBViewer?
I fixed a bug yesterday, perhaps it relate to the compatibility problem.

[sneaker_ger]

Ok, but do you observed the output of frames? Do the MPC-HC keep black screen forever?

It will only output black frames (though with the correct width & height it seems). It would sometimes output 1 or 2 non-black frames in older versions, but I cannot reproduce it anymore.

[xv]

If the win32.exe works fine, I believe the frame order is OK. The sequence is one of my test sequences, I often use it to verify the correctness of my decoder.
I resized/compressed the output of your decoder and attached it to this post. During the zoom it jumps back several times. With other files I decoded that didn´t happen, but I only tested progressive files.

I fixed a bug yesterday, perhaps it relate to the compatibility problem.Great! Could you send me a trial version to check that? I´ll already send you a mail. If it works I´ll buy DiAVC immediately.

[schweinsz]

It will only output black frames (though with the correct width & height it seems). It would sometimes output 1 or 2 non-black frames in older versions, but I cannot reproduce it anymore.
Which splitter do you use?

[sneaker_ger]

mkv: haali, gabest
mp4: gabest
ts: gabest

[Guest]

@schweinsz

What splitter(s) do you recommend?

[schweinsz]

@schweinsz

What splitter(s) do you recommend?
I use the Haali and gabest, but I can't get the question.

[Guest]

I just meant which ones do you think work best with your decoder.

You have apparently answered the question. Thank you.

[schweinsz]

I resized/compressed the output of your decoder and attached it to this post. During the zoom it jumps back several times. With other files I decoded that didn´t happen, but I only tested progressive files.
I verified the output of the DiAVC cmd with the jm just then and found that it IS correct. The sequence is generated using the JM by me to verify the correctness of my decoder, but I don't know why it is so bad.

[DeathWolf]

Download the new built DiAVC filter (http://di-avc.com/diavc.7z), try it with the UAC opening. I built the new version with the "UAC Execution Level"==RequireAdministrator. Perhaps it helps.

I tried this one, but I only get black frames in mpc-hc, regardless of the splitter.

stage0/stage1/diavc init appear first, then the frame resizes, and after that start receive, but no matter how many times I press ok I never get any actual valid non black frame.

EDIT: forgot to mention, core i7 on windows 7 x64(but 32 bit player)
EDIT2: oh and UAC is off, so no UAC issues

[TheRyuu]

I tried this one, but I only get black frames in mpc-hc, regardless of the splitter.

stage0/stage1/diavc init appear first, then the frame resizes, and after that start receive, but no matter how many times I press ok I never get any actual valid non black frame.

EDIT: forgot to mention, core i7 on windows 7 x64(but 32 bit player)
EDIT2: oh and UAC is off, so no UAC issues

Same boat as DW.

On Windows 7 64bit and DiAVC is crashing (with 32bit mpc). I also have UAC off and on the same setup as DW as well.

[Fadeout]

The filter hasn't been updated in months, so we have to wait at the very least to try a new version.

[paulvdb]

I just tested the x64 diavcx64v1.7z. It works on Windows 7 x64 but crashes on Windows XP x64. It goes as far as this:
nHighestFeature 11
nLogicalProcessors 32
intelcpu 1 corenum 32 multithrd 1 decoders 16 inited!
1 decoder inited!
2 decoder inited!
3 decoder inited!
4 decoder inited!
5 decoder inited!
6 decoder inited!
7 decoder inited!
8 decoder inited!
9 decoder inited!
10 decoder inited!
11 decoder inited!
12 decoder inited!
13 decoder inited!
14 decoder inited!
15 decoder inited!
And then it crashes.

[schweinsz]

I just tested the x64 diavcx64v1.7z. It works on Windows 7 x64 but crashes on Windows XP x64. It goes as far as this:
nHighestFeature 11
nLogicalProcessors 32
intelcpu 1 corenum 32 multithrd 1 decoders 16 inited!
1 decoder inited!
2 decoder inited!
3 decoder inited!
4 decoder inited!
5 decoder inited!
6 decoder inited!
7 decoder inited!
8 decoder inited!
9 decoder inited!
10 decoder inited!
11 decoder inited!
12 decoder inited!
13 decoder inited!
14 decoder inited!
15 decoder inited!
And then it crashes.
I am curious that the __cpuid get the wrong number of the logical processors on XP x64.
Ok, download the new DiAVC x64 cmd (http://www.di-avc.com/diavcx64v2.7z) and try it.

[TheRyuu]

I am curious that the __cpuid get the wrong number of the logical processors on XP x64.
Ok, download the new DiAVC x64 cmd (http://www.di-avc.com/diavcx64v2.7z) and try it.

I'm on Windows 7 64bit and that new one says I have 16 logical processors (I have an i7 with HT on) so shouldn't it say 8?

Edit:
diavcx64>win32.exe tempete_704x480.264 tempe
te_704x480.yuv
nHighestFeature 11
nLogicalProcessors 16
intelcpu 1 corenum 16 multithrd 1 decoders 14 inited!
1 decoder inited!
2 decoder inited!
3 decoder inited!
4 decoder inited!
5 decoder inited!
6 decoder inited!
7 decoder inited!
8 decoder inited!
9 decoder inited!
10 decoder inited!
11 decoder inited!
12 decoder inited!
13 decoder inited!
249 frames decoded totally.
68138 counters used by decoder and 461222 counters used by others.

Decoding speed: 12923 fps

D:\Encoding\diavcx64v2\diavcx64>pause

[Stephen R. Savage]

[...]\diavcx64>win32.exe tempete_704x480.264 te
mpete_704x480.yuv
nHighestFeature 10
nLogicalProcessors 2
intelcpu 1 corenum 2 multithrd 1 decoders 5 inited!
1 decoder inited!
2 decoder inited!
3 decoder inited!
4 decoder inited!
249 frames decoded totally.
609686 counters used by decoder and 1018987 counters used by others.

Decoding speed: 795 fps

[...]\diavcx64>pause
Press any key to continue . . .

Windows 7 64-bit, Core 2 Duo. I am a bit curious as to why it chooses to start 5 threads. Incidentally, DiAVC 1.1 works fine in MPC-HC 32-bit for me.

[paulvdb]

I am curious that the __cpuid get the wrong number of the logical processors on XP x64.
Ok, download the new DiAVC x64 cmd (http://www.di-avc.com/diavcx64v2.7z) and try it.

nHighestFeature 11
nLogicalProcessors 32
intelcpu 1 corenum 32 multithrd 1 decoders 14 inited!
1 decoder inited!
2 decoder inited!
3 decoder inited!
4 decoder inited!
5 decoder inited!
6 decoder inited!
7 decoder inited!
8 decoder inited!
9 decoder inited!
10 decoder inited!
11 decoder inited!
12 decoder inited!
13 decoder inited!
249 frames decoded totally.
55817866 counters used by decoder and 562498740 counters used by others.

Decoding speed: 14903 fps

[TheRyuu]

So what's the eta on fixing this for Windows 7 64bit?

[schweinsz]

So what's the eta on fixing this for Windows 7 64bit?
I meet some trouble that I am not familar with, so I need several days to fix it.

[Emulgator]

Maybe this helps when bugfixing:
While having DiAVC 1.1.1 32bit working fine on a WinXPSP3,
another System of mine (Win7U64bit using MediaPlayerClassic 32bit, ffdshow 3305 32 bit
(+ ffdshow 64 bit for Veedub64/Avisynth64)
seems unable to use DiAVC 1.1.1 32bit.
After running the installer DiAVC is registered, but with a merit below ffdshow.
A: I used Win7DS FilterTweaker 3.3 to give DiAVC the "preferred" flag, but still MPC calls ffdshow for .264.
B: Manually setting "External filters" in MPC is not respected at all.
(Well, MPC was not made for Win7). Uses ffdshow here, even if is is blacklisted.
C: I found DiAVC Setting.exe 1.0.0.1 within this thread. (schweinsz, please make it "Stickie" in the first post!)
Ran this as Administrator, gave DiAVC a higher merit than ffdshow,
hit "Change Merit", "Set DefaultH.264 Decoder", but again, ffdshow is preferred by MPC.
When I call DiAVCSetting.exe again, for DiAVC the same lower merit from before (-4194303) comes up.
ffdshow is listed as having merit -8388607.
(How comes the merit is negative and 7-figured ? Wrong interpretation as signed integer ?
Should be 8-figured and positive)

[MythCreator]

On my Win7U64,DiAVC is just crashed.I tried MPC-HC,KMP,PotPlayer & MPlayer（32bit edition)，only MPlayer works fine,the others crashed immediately

[TheRyuu]

I meet some trouble that I am not familar with, so I need several days to fix it.

So what's the current status on the fix? When can we expect an update?

[Superb]

@schweinsz: you should really start using VirtualBox to help yourself debug 64bits installations.

[NikosD]

Hello Schweinsz.
I'm trying to check out your bold statement "World's Fastest H.264 High-Definition Software Decoder"
I saw the comparison charts at your site and I think FFMpeg-mt is missing.
FFMpeg-mt is very fast nowdays and most of the times, faster than CoreAVC and DivX H.264 you publish at your site.
The fastest software decoder I know is Splash Pro by Mirillis.
Have you tried it?

How can I test your decoder ? The demo has expired.

I have a C2D-E7300 - Win 7 x64 system

Thank you

[Episodio1]

NIkosD, you know it's fastest cause you have made tests. Could you post your results?

[NikosD]

It's the only decoder capable of 60fps - steady framerate - in my system with the clip below - L4.2 1080p60fps

Sample:
http://rapidshare.com/files/415777315/ken_ross_02-1920x1080_60fpsRef4.mts.html

I couldn't try DiAVC

[woah!]

It's the only decoder capable of 60fps - steady framerate - in my system with the clip below - L4.2 1080p60fps

Sample:
http://rapidshare.com/files/415777315/ken_ross_02-1920x1080_60fpsRef4.mts.html

I couldn't try DiAVC

i have a video here at 4096 x 2304 and it is the only decoder i have that can play it smoothly at 24fps...

[Stephen R. Savage]

Claiming that ffmpeg is faster than Core, DivX, or DiAVC is absurd. For example, on a 720p video I have:

CoreAVC: 135 fps
DiAVC: 160 fps
ffmpeg-mt (ffdshow): 120 fps

[bob0r]

Maybe we should ask the MPEG-LA who may claim to be the fastest!

[NikosD]

Claiming that ffmpeg is faster than Core, DivX, or DiAVC is absurd. For example, on a 720p video I have:

CoreAVC: 135 fps
DiAVC: 160 fps
ffmpeg-mt (ffdshow): 120 fps

You have to be more careful with your answers and with mine, too

I wrote "most of the times", not always.

So it depends on the clip and the CPU architecture.

DivX, CoreAVC, FFMpeg-mt are very close in terms of performance.

Splash Pro is faster than DivX, CoreAVC, FFMpeg-mt

DiAVC, I would like to test it.

[schweinsz]

Hello Schweinsz.
I'm trying to check out your bold statement "World's Fastest H.264 High-Definition Software Decoder"
I saw the comparison charts at your site and I think FFMpeg-mt is missing.
FFMpeg-mt is very fast nowdays and most of the times, faster than CoreAVC and DivX H.264 you publish at your site.
The fastest software decoder I know is Splash Pro by Mirillis.
Have you tried it?

How can I test your decoder ? The demo has expired.

I have a C2D-E7300 - Win 7 x64 system

Thank you

I ever tested the ffmpeg-mt this spring and found that is slower than the DivX and CoreAVC.
I am downloading the Splash Pro to test it.
When I fixed the current bug, I will publish a demo version and a retail version.

[NikosD]

I ever tested the ffmpeg-mt this spring and found that is slower than the DivX and CoreAVC.
I am downloading the Splash Pro to test it.
When I fixed the current bug, I will publish a demo version and a retail version.

The easiest way to check out FFMpeg-mt, is by downloading FFDshow, because FFMpeg-mt is included.
The project is being updated a lot, so the spring version is probably outdated.
Looking forward to your Splash tests and DiAVC trial software of course.

[schweinsz]

The easiest way to check out FFMpeg-mt, is by downloading FFDshow, because FFMpeg-mt is included.
The project is being updated a lot, so the spring version is probably outdated.
I will download a new ffdshow to test it.

Looking forward to your Splash tests and DiAVC trial software of course.
It seems that the Splash AVC decoder is not a directshow-based solution so I can't test it using the timecodec or graphstudio.
I use the mirillis Splash Pro to play the "gold curse2.mkv", the one of my test set on Di-AVC.com, (I ever uploaded it (http://www.multiupload.com/1I3WQD3YVK) to the multiupload , but I can't open the page now). I disable the deinterlace and hardware-accelerated video decoding, and find the playback was very sluggish and buggy. I saw green screen or strip and I am surprised that the CPU fullness is about 85%. While the DiAVC is used the playback is fluent but the CPU fullness is 95%. I hesitate that if the option "diaable hardware-accelerated video decoding" makes sense.

[NikosD]

I will download a new ffdshow to test it.


It seems that the Splash AVC decoder is not a directshow-based solution so I can't test it using the timecodec or graphstudio.
I use the mirillis Splash Pro to play the "gold curse2.mkv", the one of my test set on Di-AVC.com, (I ever uploaded it (http://www.multiupload.com/1I3WQD3YVK) to the multiupload , but I can't open the page now). I disable the deinterlace and hardware-accelerated video decoding, and find the playback was very sluggish and buggy. I saw green screen or strip and I am surprised that the CPU fullness is about 85%. While the DiAVC is used the playback is fluent but the CPU fullness is 95%. I hesitate that if the option "diaable hardware-accelerated video decoding" makes sense.

Yes indeed. Splash pro is not a DS based solution.
It is to diffcult to test it like the other decoders.
BTW, I don't use timecodec, graphstudio is way better.

But come on! Try another clip, not the one with green stripes. You can do better than this, I'm sure.

I would be grateful if you gave me a chance to try your decoder.

P.S I don't have any special preference to Splash Pro. The same is true for any other decoder. I keep the same distance from all.

[schweinsz]

Yes indeed. Splash pro is not a DS based solution.
It is to diffcult to test it like the other decoders.
BTW, I don't use timecodec, graphstudio is way better.

But come on! Try another clip, not the one with green stripes. You can do better than this, I'm sure.

I tried the kelly love fights.mp4 just now, and found the playback is still sluggish and seemed that many frames are dropped. But the CPU fullness is about 80%.

[NikosD]

I tried the kelly love fights.mp4 just now, and found the playback is still sluggish and seemed that many frames are dropped. But the CPU fullness is about 80%.

Run FRAPS. Is the framerate steady ?

[schweinsz]

Run FRAPS. Is the framerate steady ?
I downloaded the FRAPS to benchmark the playback of the kelly love fights.mp4 using the Splash Pro, the result is as follows, in my humble word, it is vvvvvvvery slow.
FPS
17
15
17
17
14
19
15
10
15
16
17
17
14
16
13
13
8
14
6
16
18
17
17
16
12
13
12
15
16
16
14
8
15
7
16
18
18
14
13
12
12
12
16
15
15
10
13
16
14
14
12
14
14
13
15
13
13
12
13
16
16
17
10
11
6
14
15
17
15
11
18
9
16
11
12
14
12
15
13
15
18
12
11
7
11
7
17
18
18
15
17
12
9
9
15
15
11
12
11
10
14
11
12
10
10
11
14
16
16
16
11
12
9
15
10
14
15
12
9
11
16
16
8
14
9
19
19
12
9
17
11
16
18
14
9
17
16
14
14
16
16
8
13
10
7
11
11
9
18
11
17
17
18
16
17
18
18
19
9
17
18
17
17
18
17
17
9
18
11
18
16
17
17
17
17
18
17
17
17
19
10
16
18
18
17
17
18
18
12
15
17
17
18
17
19
18
18
16
16
18
16
18
17
16
16
17
19
17
10
17
17

[NikosD]

Even though I have a dual core system like yours, your first clip I tried was too easy for me.
The sample provided by me, Splash pro was and is the ONLY one played it at 60fps steady.
Your core duo has only one SSE2 unit, maybe this is the reason. Core 2 Duo has 3 units.
We need more tests by more CPU architectures
All the feedback by Splash Pro users, is that it is an extremely fast video player.

I'm waiting for your software to give it a try

[Disabled]

what is your GPU NikosD? Because perhaps Splash Pro is using GPU acceleration for you and everything else is not.

[schweinsz]

Even though I have a dual core system like yours, your first clip I tried was too easy for me.
The sample provided by me, Splash pro was and is the ONLY one played it at 60fps steady.
Your core duo has only one SSE2 unit, maybe this is the reason. Core 2 Duo has 3 units.
We need more tests by more CPU architectures
All the feedback by Splash Pro users, is that it is an extremely fast video player.
I'm waiting for your software to give it a try
The SSSE3 helps some, say 10% in DiAVC, the SSE3,SSE4 help little. The thing that DO matters is your graphics card.

[NikosD]

what is your GPU NikosD? Because perhaps Splash Pro is using GPU acceleration for you and everything else is not.

It has an option to select or deselect GPU.
I have an ATI card 5750 - UVD 2.2 - and is not getting in UVD mode.

The funny thing is, that when I select the GPU mode (DXVA), the clip plays at 60fps for a few seconds and then goes to 27-30fps.

So, it's definitely CPU mode only. It's faster than GPU in this clip, it seems that 60fps is hard for DXVA for the most DXVA players.

Only WMP12 and PowerDVD 10 are capable of DXVA H.264 1080p60FPS nowdays.

Like YouTube - UHD videos (4096 x 2304), the H.264 1080p60FPS clips definitely need fast CPU decoders, because DXVA has a lot of problems with those files. Black screens, crashing of applications etc (YouTube - 4K)

[NikosD]

The SSSE3 helps some, say 10% in DiAVC, the SSE3,SSE4 help little. The thing that DO matters is your graphics card.

No, this is irrelevant.

I didn't write about SIMD instruction sets. As a matter of fact your Core Duo has SSE3 and SSSE3 instructions sets.

If you read it carefully, I wrote about the implementation of SIMD architecture inside the CPU. This has nothing to do with instruction sets

[NikosD]

I downloaded the FRAPS to benchmark the playback of the kelly love fights.mp4 using the Splash Pro, the result is as follows, in my humble word, it is vvvvvvvery slow.


I think I have a clue of this strange behavior.
You have to go to the Playback options window- the left button of the first screen - and disable Best quality, Motion2, Detail boost and all the other added processing.

Maybe that is your problem with slow performance.

Because you are the only one I know, who is complaining about slow performance of this particular player.

Give it a try.

[schweinsz]

I think I have a clue of this strange behavior.
You have to go to the Playback options window- the left button of the first screen - and disable Best quality, Motion2, Detail boost and all the other added processing.

Maybe that is your problem with slow performance.

Because you are the only one I know, who is complaining about slow performance of this particular player.

Give it a try.
I can't find the "Best quality", "Motion2" and "Detail boost", I only disable the deinterlacing and hardware accelerating. I only have the trial version of the Splash player.

[schweinsz]

No, this is irrelevant.

I didn't write about SIMD instruction sets. As a matter of fact your Core Duo has SSE3 and SSSE3 instructions sets.

If you read it carefully, I wrote about the implementation of SIMD architecture inside the CPU. This has nothing to do with instruction sets

The intel architecture follows same rules. If a decoder is slow on Core Duo, the same factor follows on other platforms such as Core 2 Duo, Core i3, i7, etc. in most cases. Besides, my Core Duo has not the SSSE3.

[NikosD]

The intel architecture follows same rules. If a decoder is slow on Core Duo, the same factor follows on other platforms such as Core 2 Duo, Core i3, i7, etc. in most cases. Besides, my Core Duo has not the SSSE3.

Not exactly, I think.
Your code I suppose is hand made assembly, so it is different to code for 3 sse2 units and different for 1.
And it sure makes a lot of difference to Intel Compiler.

[NikosD]

I can't find the "Best quality", "Motion2" and "Detail boost", I only disable the deinterlacing and hardware accelerating. I only have the trial version of the Splash player.

You have to click the left little sign at the first panel with play/ stop button.
It doesn't write anything. It has just a small sign.

[Mixer73]

You have to click the left little sign at the first panel with play/ stop button. It doesn't write anything. It has just a small sign.

Honestly seeing as you're the one here shilling this product why don't you present some empirical evidence of the superiority of Splash Player? 1 clip of 1 resolution on one test PC does not a comparison make.

[NikosD]

Honestly seeing as you're the one here shilling this product why don't you present some empirical evidence of the superiority of Splash Player? 1 clip of 1 resolution on one test PC does not a comparison make.

To be honest, I have no profit by Splash or any other company selling video players as I have written to a previous post.
That means I have no profit by DiAVC either.
It looks like Schweinsz made me a Splash supporter, because of the directions I had to give for equal benchmarking.
As I said to my first post, it is a bold statement to say "World's Fastest H.264 High-Definition Software Decoder"
That is what I'm trying to check out. Maybe it is, maybe it is not.
I don't have DiAVC yet, I'm waiting for a working version, because whatevet I tried from the site, it didn't work.
And of course is not only one clip. It's the most difficult clip for my system, capable to produce clear differences to video players.

[Episodio1]

No benchmarks on official Splash website? Weird... :(

[Mixer73]

It looks like Schweinsz made me a Splash supporter, because of the directions I had to give for equal benchmarking. As I said to my first post, it is a bold statement to say "World's Fastest H.264 High-Definition Software Decoder"

There's n+1 tests here by Schweinz and many others proving that DiAVC is very competitive, over lots of types of files and system configuration. No claim of "fastest" will ever stand up to every test you throw at it, each will have its strengths and its weaknesses.

Claiming that this other product is faster while providing zero statistics at all, while staking your claim to doubt Schweinz's claim, and only having tested one clip on one system is quite honestly laughable.

[NikosD]

There's n+1 tests here by Schweinz and many others proving that DiAVC is very competitive, over lots of types of files and system configuration. No claim of "fastest" will ever stand up to every test you throw at it, each will have its strengths and its weaknesses.

Claiming that this other product is faster while providing zero statistics at all, while staking your claim to doubt Schweinz's claim, and only having tested one clip on one system is quite honestly laughable.

Quite honestly laughable is that DiAVC is not testable.
Because it does not exist right now. So, there is no debate.

BTW, PotPlayer uses FFMpeg-mt in a faster way than FFDshow and it IS FASTER than Splash in that way.

I have a question though about its quality.

[Superb]

@NikosD:
Oh, stop whining. You can buy it and test it, if you really want to.
He didn't write "World's Fastest FREE H.264 High-Definition Software Decoder".
He also said he'll provide a future (working) demo version.
I've seen many real benchmarks in this thread (numbers! from different people w/ many types of machines!), and each and every one says the exact same thing:
DiAVC is indeed the world's fastest H.264 software decoder.

You writing "A is faster" or "B is faster" w/o providing any benchmarks is simply useless.

Btw, you cannot question the quality of a H.264 decoder.
All decoders are supposed to give you the exact same output (pixel to pixel).
If one decoder doesn't, then it has a bug.

[Underground78]

Quite honestly laughable is that DiAVC is not testable.

Well, people like me who have bought DiAVC can tell you it is fast ...

[NikosD]

@NikosD:
Oh, stop whining. You can buy it and test it, if you really want to.
He didn't write "World's Fastest FREE H.264 High-Definition Software Decoder".
He also said he'll provide a future (working) demo version.
I've seen many real benchmarks in this thread (numbers! from different people w/ many types of machines!), and each and every one says the exact same thing:
DiAVC is indeed the world's fastest H.264 software decoder.

You writing "A is faster" or "B is faster" w/o providing any benchmarks is simply useless.

Btw, you cannot question the quality of a H.264 decoder.
All decoders are supposed to give you the exact same output (pixel to pixel).
If one decoder doesn't, then it has a bug.

I'm not whinning. It's not my style.

Buy it if you want to test it. This is not a nice way to sell your product. That's why all the developers who want to sell their products have a trial demo.

I have never said which is the fastest decoder/player.
I have only said try this and that because there are no numbers for this and that.
FFMpeg-mt, Splash and PotPlayer are sometimes faster than DivX/ CoreAVC and are not included in your comparison charts at your site. Maybe they are faster than your product, too.

You have to be more careful reading my answers. Non DS decoders like Splash and PotPlayer can not be benchmarked by tools like graphstudio.

So, from all the above IMHO, you have to rephrase your moto to a more decent phrase like this:
"One of the World's Fastest H.264 High-Definition Software Decoders"

This is the last post by me, for the fastest-thing debate.

[Underground78]

I can't test non DS decoders but according to my tests DiAVC is faster than ffmpeg-mt and DivX H264 Decodeur :

DiAVC :
User: 0s, kernel: 0s, total: 0s, real: 11s, fps: 651.9, dfps: 47.0

ffdshow :
User: 27s, kernel: 0s, total: 27s, real: 28s, fps: 19.7, dfps: 19.5

ffmpeg-mt :
User: 1s, kernel: 0s, total: 1s, real: 16s, fps: 510.1, dfps: 33.5

DivX H264 Decoder :
User: 1s, kernel: 0s, total: 1s, real: 14s, fps: 495.8, dfps: 39.3

I have no reason to lie about the results and it seems to confirm schweinsz's tests ... I don't see how to test non DS decoders but if you have a proposal to do so, I can try ...

[NikosD]

You didn't read my previous answers to Schweinsz.
AFAIK the only way is FRAPS.

About the quality vs performance issue, you CAN cheat by skipping deblocking or altering other inside parameters hidden in All-in-players and not customizable by user.

[Underground78]

You didn't read my previous answers to Schweinsz.
AFAIK the only way is FRAPS.

Yep, I read that but Schweinsz's test using Fraps did not seem to work very well, it was quite slow so if you think it's the correct way to benchmark it, it seems it is not faster than DiAVC at all ...

About the quality vs performance issue, you CAN cheat by skipping deblocking or altering other inside parameters hidden in All-in-players and not customizable by user.

Well, if it was the case I am sure someone would have noticed this ... I made comparison with DGAVCDecNV and DGAVCDecDi (which is the only reason I have bought DiAVC, ffdshow-mt and DivX H264 are fast enough for daily use and free ...) lately and both were identical.

[NikosD]

Well, people like me who have bought DiAVC can tell you it is fast ...

Give me a good reason to buy such product - a software H.264 decoder - when:

1) There is FFMpeg-mt which is a free software H.264 decoder
2) Most of CPU processors are at least dual core
3) There are free DXVA players/codecs (PotPlayer, MPC-HC, FFDShow DXVA) plus WMP12 for Windows 7

[Underground78]

Give me a good reason to buy such product - a software H.264 decoder - when:

1) There is FFMpeg-mt which is a free software H.264 decoder
2) Most of CPU processors are at least dual core
3) There are free DXVA players/codecs (PotPlayer, MPC-HC, FFDShow DXVA) plus WMP12 for Windows 7

Well, I answer your question before you ask so see last post, short answer is "DGAVCDecDi". :)

[NikosD]

The cheating goes more to All-in-one players, not the customizable codecs.

[NikosD]

Well, I answer your question before you ask so see last post, short answer is "DGAVCDecDi". :)

Yeah, I saw it after my post :)

[Fadeout]

If you are close to the limit DiAVC buys you some more space because it's faster. Same reasons for buying CoreAVC.

For a number of systems they make the difference.

[TheRyuu]

Soooooo about that Windows 7 64bit bug... :)

[schweinsz]

I uploaded a new version of the DiAVC Trial version (http://di-avc.com/diavc_trial_1.2.0.7z) to di-avc.com. Please download it and use it on win7 x64, perhaps it works.

[TheRyuu]

I uploaded a new version of the DiAVC Trial version (http://di-avc.com/diavc_trial_1.2.0.7z) to di-avc.com. Please download it and use it on win7 x64, perhaps it works.

It does indeed work.

[Guest]

Congratulations, schweinsz, and thank you.

[sneaker_ger]

Yes, I did a very short test and it seems to work indeed.

[NikosD]

Yes, it works.

And yes it's the slowest.

System:
Win 7 x64 - C2D@2.83GHz

Sample Test:
http://rapidshare.com/files/416586142/1.Twinpeaks1080p30fpsRef2-27Mbps.mov.html

Using graphstudio (3 passes) average results:

CoreAVC v2.0
Work time: 00:00:07:656 107.8842 fps
FFMpeg-mt rev 3563
Work time: 00:00:08203 100.6872 fps
DiAVC v1.2.0.7
Work time 00:00:08.316 99.3296 fps

Third out of three for the "World's Fastest H.264 High-Definition Software Decoder"

[schweinsz]

Yes, it works.

And yes it's the slowest.

System:
Win 7 x64 - C2D@2.83GHz

Sample Test:
http://rapidshare.com/files/416586142/1.Twinpeaks1080p30fpsRef2-27Mbps.mov.html

Using graphstudio (3 passes) average results:

CoreAVC v2.0
Work time: 00:00:07:656 107.8842 fps
FFMpeg-mt rev 3563
Work time: 00:00:08203 100.6872 fps
DiAVC v1.2.0.7
Work time 00:00:08.316 99.3296 fps

Third out of three for the "World's Fastest H.264 High-Definition Software Decoder"
This is movie trailer that are all coded using the CAVLC. But most of the movies available are coded by CABAC. You should test the decoders using more bitstreams.

[TheRyuu]

It's faster on everything I've tested so far.
Using dss2 and avs2avi (-o n -c null) on the rat.264 file (muxed to mkv) in the dg benchmark thread (this is on an i7@3.8ghz):

ffmpeg-mt: ~209fps
diavc: ~270fps

And I'm fairly certain people would mainly use this for cabac decoding, which it should (and appears to) be the fastest in the vast majority of cases.

[NikosD]

This is movie trailer that are all coded using the CAVLC. But most of the movies available are coded by CABAC. You should test the decoders using more bitstreams.

OK, let's play by your rules.

New clip. A CABAC one this time.

Here:
http://rapidshare.com/files/417906037/Paranormal_Activity.mkv.html

DiAVC is still the slowest

CoreAVC
Work time: 00:00:24.987 102.8126 fps
FFMpeg-mt
Work time: 00:00:25.445 100.9624 fps
DiAVC
Work time: 00:00:26.441 97.1589 fps

Third out of three.

It makes me wonder what will be your next excuse to justify the "World's Fastest H.264 High-Definition Software Decoder" title

OK let's be honest to each other and the others.
Change the title to something more decent and modest for your potential customers.

[Disabled]

How are you measuring the performance? There are wrong ways to measure this you know? (I'd never thing I'd bring the CoreAVC excuse here...)
I have a similar system and DiAVC always was the fastest for me...
Those are all 32Bit codecs?

[madshi]

@NikosD, I don't see you complaining in the CoreAVC thread. They also claim to have the fastest software decoder. Why are you only complaining in this thread and not in the CoreAVC thread? Do you actually think that your benchmarks are the only ones that count and that all benchmarks done by other users are worthless? It seems that many people report that DiAVC is fastest for them. You seem to have better results with CoreAVC. That's fine, but if you want to complain, at least be fair and complain in both threads.

[Underground78]

DiAVC is clearly not optimized for CAVLC (but maybe schweinsz can do something for this ?) but I have different results than yours for your CABAC sample :
FFMpeg-mt : 80.8 fps
DivX H264 Decoder : 99.0 fps
CoreAVC : 105.9 fps
DiAVC : 109.5 fps

(Tested on Windows XP SP3 with an Athlon X2 6000+)

Edit : on my laptop :

CoreAVC (cuda) : 63.0 fps
FFMpeg-mt : 83.5 fps
CoreAVC : 86.5 fps
DiAVC : 104.8 fps

Don't know why CoreAVC is so slow here.

[NikosD]

@NikosD, I don't see you complaining in the CoreAVC thread. They also claim to have the fastest software decoder. Why are you only complaining in this thread and not in the CoreAVC thread? Do you actually think that your benchmarks are the only ones that count and that all benchmarks done by other users are worthless? It seems that many people report that DiAVC is fastest for them. You seem to have better results with CoreAVC. That's fine, but if you want to complain, at least be fair and complain in both threads.

Complain with others results ?

This is the best advice I have heard for a long time.

I have aleady said that FFMpeg-mt, DivX, CoreAVC and now after my tests, DiAVC are very close in terms of performance generally.
There is no clear winner.
It depends on the clip and the CPU architecture.

My advice:
Use free software decoders like FFMpeg-mt and free DXVA decoders like FFDShow DXVA, MPC-HC, PotPlayer

@Superb and Mixer73
Where are you hiding ?

[madshi]

I have aleady said that FFMpeg-mt, DivX, CoreAVC and now after my tests, DiAVC are very close in terms of performance generally.
There is no clear winner.
It depends on the clip and the CPU architecture.
In that case neither DiAVC nor CoreAVC should claim to have the fastest decoder, right? So again: Why do you complain about DiAVC's claim, and why do you not complain about CoreAVC's similar claim?

My advice:
Use free software decoders like FFMpeg-mt and free DXVA decoders like FFDShow DXVA, MPC-HC, PotPlayer
Are you aware that it is quite rude to enter a thread and then post "my advice is to use something else"?

Anyway, you've posted your opinion, you've posted your advice, so maybe now it's time for you to move on and leave this thread be. Thank you.

[Underground78]

@schweinsz : are you aware that DiAVC can play Lossless H264 files created using x264 ?

[NikosD]

In that case neither DiAVC nor CoreAVC should claim to have the fastest decoder, right? So again: Why do you complain about DiAVC's claim, and why do you not complain about CoreAVC's similar claim?


Are you aware that it is quite rude to enter a thread and then post "my advice is to use something else"?

Anyway, you've posted your opinion, you've posted your advice, so maybe now it's time for you to move on and leave this thread be. Thank you.

I gave advice because you gave advice. Complain to another thread etc etc

To your next advice I reply that I have no intention to close this thread. Nor the power to close it, even if you seem to give me such power.

And I can write to every thread I want, whenever I want.

Keep these kind of advices to yourself and others.

Thank you

[NikosD]

@Schweinsz

Keep up the good work.

Competition is good and necessary.

[sneaker_ger]

Core i7-860:

CAVLC:
DiAVC: 348 fps
ffdshow (mt): 315 fps

CABAC:
DiAVC: 452 fps
ffdshow (mt): 326 fps

[schweinsz]

@schweinsz : are you aware that DiAVC can play Lossless H264 files created using x264 ?
Currently no.

[schweinsz]

DiAVC is clearly not optimized for CAVLC (but maybe schweinsz can do something for this ?) but I have different results than yours for your CABAC sample :
FFMpeg-mt : 80.8 fps
DivX H264 Decoder : 99.0 fps
CoreAVC : 105.9 fps
DiAVC : 109.5 fps

(Tested on Windows XP SP3 with an Athlon X2 6000+)

Edit : on my laptop :

CoreAVC (cuda) : 63.0 fps
FFMpeg-mt : 83.5 fps
CoreAVC : 86.5 fps
DiAVC : 104.8 fps

Don't know why CoreAVC is so slow here.

The DiAVC is highly optimized using the SSSE3, so I found that the DiAVC has a big advantage over other software H.264 decoder on my test when using the CPU with SSSE3. I surpose that your laptop is with a SSSE3 CPU.

[schweinsz]

@Schweinsz

Keep up the good work.

Competition is good and necessary.

I will download your two clips and test it on my two computers. It is very slow to download the second clip.

[Underground78]

Currently no.

I've missed a negation, I wanted to write : "are you aware that DiAVC cannot play Lossless H264 files created using x264 ?". But you seem to know that.

The DiAVC is highly optimized using the SSSE3, so I found that the DiAVC has a big advantage over other software H.264 decoder on my test when using the CPU with SSSE3. I surpose that your laptop is with a SSSE3 CPU.

It's an Intel Mobile Core 2 Duo P7350 and it seems you are right, it does support SSSE3.

[TheRyuu]

CoreAVC v2.0
Work time: 00:00:07:656 107.8842 fps
FFMpeg-mt rev 3563
Work time: 00:00:08203 100.6872 fps
DiAVC v1.2.0.7
Work time 00:00:08.316 99.3296 fps

Strange I cannot replicate your results although I think I know why.
For the twinpeaks:
diavc:
411.22fps
ffmpeg-mt:
365.21fps
(divx would crash when I tried to dss2 it)

CoreAVC
Work time: 00:00:24.987 102.8126 fps
FFMpeg-mt
Work time: 00:00:25.445 100.9624 fps
DiAVC
Work time: 00:00:26.441 97.1589 fps
And for paranormal:
diavc:
488.74fps
divx:
427.50fps
ffmpeg-mt:
394.50fps

Because you're only on a core2duo diavc can't stretch it's muscle and make use of how much better it can thread than ffmpeg-mt (my theory at least).

I tried setting both ffmpeg-mt and diavc to only use 2 threads and sure enough on twinpeaks they both gave a similar result of about 183fps.

Give it enough threads and diavc will beat anything out there. I do agree with you that diavc is probably a niche market item (dgavcdecdi comes to mind) otherwise ffmpeg(-mt) is plenty fast enough for just playback on any modern system.

[yesgrey]

The fastest software decoder I know is Splash Pro by Mirillis.

Splash Pro is faster than DivX, CoreAVC, FFMpeg-mt

CoreAVC v2.0
Work time: 00:00:07:656 107.8842 fps
FFMpeg-mt rev 3563
Work time: 00:00:08203 100.6872 fps
DiAVC v1.2.0.7
Work time 00:00:08.316 99.3296 fps
CoreAVC
Work time: 00:00:24.987 102.8126 fps
FFMpeg-mt
Work time: 00:00:25.445 100.9624 fps
DiAVC
Work time: 00:00:26.441 97.1589 fps
It makes me wonder why you did not include the super-duper-ultra fast Splash in your testing...:rolleyes:

[sneaker_ger]

It makes me wonder why you did not include the super-duper-ultra fast Splash in your testing...:rolleyes:

I think that Splash does not offer any DirectShow filter, so you can't compare it to other decoders using graphstudio or timecodec.

[NikosD]

I think that Splash does not offer any DirectShow filter, so you can't compare it to other decoders using graphstudio or timecodec.


That's right.

[yesgrey]

That's right.
So how can you claim that Splash is the fastest without any benchmark numbers?

You were saying that you could not test it because there were no DiAVC trial available. Now the trial exists, so we are still waiting for your results...

[ExSport]

NikosD:
My tests are similar as others = DiAVC is fastest:devil:
Tested PARANORMAL sample on Core2Duo T9400 (2.53GHz) [Windows Vista 32bit]
DiAVC
real: 21s, fps: 534.7, dfps: 118.4
CoreAVC
real: 26s, fps: 550.8, dfps: 98.5
FFMPEG_MT
real: 26s, fps: 511.4, dfps: 97.8

EDIT:
And now CAVLC sample (Twinpeaks):
DiAVC
real: 8s, fps: 534.8, dfps: 101.2
CoreAVC
real: 7s, fps: 529.5, dfps: 105.9
FFMPEG_MT
real: 8s, fps: 569.3, dfps: 98.2
This sample is the first one where DiAVC is not fastest decoder what I ever seen in my tests.

[NikosD]

@ To all

The tests were done by deselecting the option "Drop non-reference frames when hard"

This option is selected by default.
I want no dropped frames when hard or when soft.

You?

[Disabled]

Can you please answer the question about how you did the test NikosD? What program did you use and what value did you read?

[NikosD]

I always use Graphstudio. What do you mean by value I read ? Average FPS after 3 passes

[paulvdb]

This new version of DiAVC works for me on Windows XP x64. Earlier versions did not work so that is the first good news. I also tested it with the function where speed is most important for me: as a source decoder for DirectShowSource in Avisynth. I take a BBC HD clip, resize to 1280x720 and encode with x264 in 2 passes. I just tested the same source with 4 decoders and got the following results (first number is fps for first pass, second number is second pass):
diav 118.46 67.19
divx 110.63 66.50
core 109.82 66.50
ffds 107.75 65.83

[NikosD]

In order to have fair results and not cheat, you have to find and deselect "Drop non-reference frames when hard".

I'm afraid that most of you have done the tests without even see this option.

Don't publish results with default selection.

[paulvdb]

When I install a codec the first thing I do is check it's property page to see what kind of options it has. In the case of DiAVC I had deselected that option before I did my tests.

[ExSport]

Same by me:
"Drop non-reference frames when hard" deselected when tested ;)

[LoRd_MuldeR]

My two cent:

Source:
http://video-test-sequences.hexagon.cc/torrents/70618-TallShip_720p60_lossless_HuffYUV

Encoder settings:
x264 core 93 r1564M a927654 / cabac=1 / ref=16 / deblock=1:-1:-1 / analyse=0x3:0x133 / me=umh / subme=10 / psy=1 / psy_rd=1.00:0.15 / mixed_ref=1 / me_range=24 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-3 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / constrained_intra=0 / bframes=8 / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / wpredb=1 / wpredp=2 / keyint=750 / keyint_min=25 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=18.0 / qcomp=0.60 / qpmin=10 / qpmax=51 / qpstep=4 / ip_ratio=1.40 / aq=1:1.00

Results:
DiAVC H.264 Decoder v1.2.0:
243.3608

DiAVC H.264 Decoder² v1.2.0:
306.2338

CoreAVC Video Decoder v1.9.5:
214.7949

DivX H.264 Decoder v9.0.1.21:
216.8876

ffdshow Video Decoder r3556:
219.2121

(Average FPS from three runs each, as reported by GraphStudio)
[2] Option "Drop non-reference frames when hard" enabled

[Mixer73]

@Superb and Mixer73
Where are you hiding ?

I'm not hiding, I never made any claims about DiAVC, but how you can you make claims when you're only tested on very limited clips on one CPU & OS when clearly there's more than 10 people here on diverse systems who's result differ from yours?

I've never bothered to benchmark DiAVC but I respect what Schweinz is doing as in independant developer, and I allow him some salesmanship in presenting his product.

Anyway I don't see benchmarks for the product you've been spruiking.

[TheRyuu]

"Drop non-reference frames when hard"
What does the option do exactly?

[LoRd_MuldeR]

"Drop non-reference frames when hard"
What does the option do exactly?

I assume that when the decoding rate of DiAVC drops below the desired playback rate, then it completely skips all frames that aren't referenced in future frames in order to catch up.

In a decoder benchmark, where the playback rate is "infinite" (as fast as possible), this will probably happen all the time. So the comparison would be unfair...

[sneaker_ger]

@ To all

The tests were done by deselecting the option "Drop non-reference frames when hard"

This option is selected by default.
I want no dropped frames when hard or when soft.

You?

I already had it deselected when I made my tests, so my numbers stay unchanged.

[Fadeout]

Still no hope for a non-trial release?

[lych_necross]

Can registered users use the new trial version to update their currently installed registered version?

[schweinsz]

Still no hope for a non-trial release?
This night.

[schweinsz]

Can registered users use the new trial version to update their currently installed registered version?
Yes, you can.

[ExSport]

Hi schweinsz
Any news on MEncoder compatibility fix?
Many thanks for info:thanks:

[Emulgator]

Congratulations, schweinsz !
diavc.ax 1.2.0 on WinXP SP3 32bit:
Now .264 elementary stream playback is fine.
(SD resolution 720x576x25i tested).
The occasional blocking artifacts on x264 MBAFF sources when played realtime in MPC are gone.

[schweinsz]

This night.
I meet some emergency things just then. I must leave the city for some days so that I can't release the version this night and can't compute the password for several days. I am very sorry for all of this. The next version will be available in one week with the x64 version and lossless support.

[schweinsz]

Congratulations, schweinsz !
diavc.ax 1.2.0 on WinXP SP3 32bit:
Now .264 elementary stream playback is fine.
(SD resolution 720x576x25i tested).
The occasional blocking artifacts on x264 MBAFF sources when played realtime in MPC are gone.

Yes, I fixed the problem in the last several days.

[Emulgator]

The same PAL-SD .264 stream on an older, but beautiful
P4 Mobile 2.0 GHz from 2003 ("Northwood", MMX, SSE, SSE2) using diavc.ax 1.2.0:
Beautifully playable, CPU load 57%..70%. This is where DiAVC is an advantage.
ffdshow 3562 could not play this stream on that particular system without stalling,
although ffdshow is almost on par with DiAVC on more recent CPUs.
Win7U64 ? Let's see...

[Guest]

@Emulgator

Can you help a DirectShow noob, please? What splitters do you use (for WinXP 32) and where do I get them? Thank you.

[xv]

I also found out that the decoder works perfect in nearly all cases: But unfortunately not in DVBViewer.

I attached two screenshots to show how it looks like, the greyscale is 1280x720 with 50fps progressive, the other one is 1920x1088 with 25fps interlaced (50 fields per second). I tested other channels, the problem is identical.

I can record these streams and play them in mpc-hc, no problem with DiAVC, the problem is not the stream, and because it also seems to be decoded correctly by DiAVC in DVBViewer (no artefacts etc.) the problem seems to be between DiAVC and the renderer. I tried different renderers supported by DVBViewer (VMR7, VMR9, Overlay), all with exactly the same result.

A x64 version and lossless support is really great, both are very useful for me! Will it support only the new lossless method or also the old one (IIRC it was removed from the H.264 spec and included in x264 until rev 988)?

[Emulgator]

neuron2: Can you help a DirectShow noob, please?
What splitters do you use (for WinXP 32) and where do I get them? Thank you.
Oops. If you say that then I must have made a mistake.

That made me go back and check what MPC builds at my DELL WinXP SP3 system T7600G CPU.
In the last few days I was experimenting with versions of diavc.ax,
their merits and and may have blown something.
I guess I unintentionally shot my sheriff...

A: Right now an extension .264 will build: nothing but a crashed MPC.
A: GraphEdit: file.264 -> ffdshow -> Videorenderer. Activating Preview stalls Graphedit.
Inserting DiAVC manuall does not work.
"The filter could not be created. Resources used by this filter may already be in use."
(Return code: 0x8007007e.)
I must have kicked DiAVC from the shelf somehow.
Or the trial version 1.2.0 has expired ? I had bought a license for version 1.1..

B: .avc (renamed from .264) with a "lower" SD bitrate (CRF18) uses Cyberlink Demuxer (from PDVD7.3) and ffdshow
(although I excluded this one in MPC for now) with MBAFF artifacts.
B: Graphedit: File.avc -> TS Input of Cyberlink Demux (PDVD7) -> ffdshow In -> Videorenderer

C: .avc (renamed from .264) with (for SD ridiculously high) 16000Mbps
uses Cyberlink Demuxer (from PDVD7.3) and ffdshow 3406.
C: Graphedit: File.avc -> TS Input of Cyberlink Demux (PDVD7) -> ffdshow In -> Videorenderer
C plays back flawlessly.
And sometimes I remember seeing that Cyberlink decoder was used too. Hm.

Now investigating what I wrote for the Toshiba P4m.
Toshiba: ffdshow Video Decoder, DiAVC Decoder. Reacts on DiAVC controls.
Toshiba Graphedit: File.264 -> XForm In DiAVC XForm Out -> ffdshow In -> Video Renderer.
This still plays ok.

P.S Never mind. I'm too tired. went to bed this morning around 8:50.
Filmerit helped me:
Forgot to rename diavc.1.1.ax to diavc.ax on the Dell.
All is working again. graph is there, plays flawlessly.

Dell A: Graphedit: File.264 -> XForm In DiAVC XForm Out -> ffdshow In -> Videorenderer

[Guest]

Still like to know about the splitter as I asked. :)

[Emulgator]

Ah, I thought you asked about .264.
(For .264 I see no splitter at my system, I expected none for a elementary stream.)

Filmerit lists the following splitters for the Dell:
Avi Splitter quartz.dll (MS)
Cyberlink MPEG Splitter clspltr.ax (PDVD7)
Cyberlink MPEG-4 Splitter clm4splt.ax (PDVD7)
DV Splitter qdv.dll (MS)
Haali media Splitter splitter.ax (3 instances, Haali)
MPEG-2 Splitter mpg2splt.ax (MS)
MPEG-I Stream Splitter quartz.dll (MS)
Nero DV Splitter nedvsplitter.ax (Nero)
Nero File Source Splitter nefsource.ax (Nero)
Nero Splitter nesplitter.ax (Nero)

Forgot Mainconcept (Adobe2) MPEG Splitter ad2mcspmpeg.ax (Premiere CS3).
This one is marked red ("not found"),
but is physically in its path position and I guess locked for exclusive use by Adobe apps.

I expected to see more about Sony (Vegas, DVD-A) and Canopus (Edius)
but these thingies are not explicitly called "Splitter".
Maybe it is a naming thing and I guess these are locked anyway to be used only by the host application.

[Guest]

Thanks. Here's what I am getting at...

Open a TS file with AVC inside in GraphStudio and show the resulting graph.

[Emulgator]

Here you go:
http://dvd-manufactur.de/files/2010 09 15 DELL GraphEdit_M2TS.png
File 00000.m2ts is from the HDMV-BD-Testdisk v 1.3b,
AVC encoded using x264 core67 r1145M 1920x1080x23.976p, L4.1, CABAC, ref=3...

And another one:
http://dvd-manufactur.de/files/2010 09 15 DELL GraphEdit_MTS.png
File 00000.mts is AVCHD encoded by a Panasonic GH-1,
Video AVC L4.0 1280x720x50p, CABAC no, ref=1...
Audio AC-3 2.0 192kbps@48kHz

[Guest]

Thanks. Hmmmm...it doesn't show what splitter is used. Do you know if it is Haali and what version?

[Midzuki]

You could try "File Source (Async)" ++ "MPC - MPEG Splitter". :)

[poisondeathray]

If you right click the 1st box, it will give you more information on the splitter. Graphstudio gives more information and options than graphedit

[Emulgator]

(back from the kitchen)
Will get GraphStudio. Moment please.

[Emulgator]

You were right, Right click on the source file-> Properties.
For the .MTS it is Haali 1.9.42.1.
For the .m2ts as well.
Filename: splitter.ax
So the GraphTools don't show the splitter by default ?
Hum, definitely worth an option.
You don't get to see it if you don't know where to tickle;-)

[Guest]

Thank you!

[DeathWolf]

Schweinzs, is there any luck you might automate a bit your activation system. Even limited number of auto activations would be better than having to email to get one.
Just an online form to get a new key for when one reinstalls or changes of computer.

[schweinsz]

Schweinzs, is there any luck you might automate a bit your activation system. Even limited number of auto activations would be better than having to email to get one.
Just an online form to get a new key for when one reinstalls or changes of computer.

Yes, I DO have a plan to fix the problem in the October.

[ExSport]

Hi schweinsz
Any news on MEncoder compatibility fix?
Many thanks for info:thanks:
Hm, tested and still not correct output with MEncoder and DiAVC as decoder :(

[fabje]

I'm using Staxrip to cut out the commercials etc out of a recording.
And now I want to use DiAVC for Directshow anyone here that can tell me how I can doe that?

[LigH]

Just ensure that DiAVC has the highest "Merit" of all installed AVC decoder filters: In Windows XP you can use e.g. Headbands' GSpot 2.70a, or the RadLight Filter Manager; and in Windows 7, you may also need to suppress Microsoft's internal "Media Foundation" decoders to re-enable DirectShow filters being used for decoding (e.g. using the "Filter Tweaker").

http://www.codecguide.com/windows7_preferred_filter_tweaker.htm

But in general ... I would not recommend to use DirectShow at all, if there are any alternatives as native AviSynth plugins available. Like DGAVCDecDI (http://neuron2.net/dgavcdecdi/dgavcdecdi.html).

[Emulgator]

schweinsz, can it be that 1.2.0 trial has expired somehow even on a licensed machine ?
Now that I want to give DiAVC a shot for working in FullHD it seems to let me down.
Source: 1920x1080x25i.
No matter if from .mts or from demuxed .264, decoding via DGSource/DGAVCIndexDI/diavc.ax,
just a decoding and splicing script, VirtualDub crashes after some 30..40 frames.
I have to work with it now, seriously.

Ah, it is a memory leak.
After emptying my base script to only 5 lines decoding to uncompressed .avi seems to work now,
but RAM usage is still increasing steadily...

And VD crashes after using up some 1.6GB RAM after writing out 7730 FullHD frames.

a=DGSource("I:\iv\2010 09 19\007.demuxed.dgi")
b=DGSource("I:\iv\2010 09 22\006.demuxed.dgi")
c=DGSource("I:\iv\2010 09 22\007.demuxed.dgi")
d=DGSource("I:\iv\2010 09 23\001.demuxed.dgi")
a ++ b ++ c ++ d

[schweinsz]

schweinsz, can it be that 1.2.0 trial has expired somehow even on a licensed machine ?
No, the current trial version 1.2.0 will expires at the end of the month. If you can playback H.264 videoes using the DiAVC 1.2.0 trial, it should work for others.

Now that I want to give DiAVC a shot for working in FullHD it seems to let me down.
Source: 1920x1080x25i.
No matter if from .mts or from demuxed .264, decoding via DGSource/DGAVCIndexDI/diavc.ax,
just a decoding and splicing script, VirtualDub crashes after some 30..40 frames.
I have to work with it now, seriously.

Ah, it is a memory leak.
After emptying my base script to only 5 lines decoding to uncompressed .avi seems to work now,
but RAM usage is still increasing steadily...

And VD crashes after using up some 1.6GB RAM after writing out 7730 FullHD frames.

a=DGSource("I:\iv\2010 09 19\007.demuxed.dgi")
b=DGSource("I:\iv\2010 09 22\006.demuxed.dgi")
c=DGSource("I:\iv\2010 09 22\007.demuxed.dgi")
d=DGSource("I:\iv\2010 09 23\001.demuxed.dgi")
a ++ b ++ c ++ d
Do you mean that the VirtualDub calls the DGSource/DGAVCIndexDI/diavc.ax?
Give me one of your e-mail, I can send you a new version to test.

[Stephen R. Savage]

schweinsz: Do you have any information on when the final version of DiAVC 1.2 will be available? I am very interested in 64-bit and lossless support, but it has been a long time since you said that you would release it.

Emulgator only means that he uses VirtualDub to open Avisynth scripts. I believe he is using the DGAVCDecDI product.

[schweinsz]

schweinsz: Do you have any information on when the final version of DiAVC 1.2 will be available? I am very interested in 64-bit and lossless support, but it has been a long time since you said that you would release it.
It is near accomplish. But I am doing some algorithms for JCTVC, so the release is delayed.

Emulgator only means that he uses VirtualDub to open Avisynth scripts. I believe he is using the DGAVCDecDI product.
OK, thanks.

[Emulgator]

Yes, that is correct.
I have used DGAVCDecDI, now swapped the delivered diavc.ax with the 1.2.0.diavc.ax.
my mail for the new trial: mail(at)dvd-manufactur.de
BTW, end of which month will 1.2.0 expire ? September ? October ?
It may be better to tell the date.

[schweinsz]

Yes, that is correct.
I have used DGAVCDecDI, now swapped the delivered diavc.ax with the 1.2.0.diavc.ax.
my mail for the new trial: mail(at)dvd-manufactur.de
BTW, end of which month will 1.2.0 expire ? September ? October ?
It may be better to tell the date.
October.

[senseiam]

Registered for DiAVC H.264 Decoder yesterday still no password, should I expect a response sometime today?

[schweinsz]

Registered for DiAVC H.264 Decoder yesterday still no password, should I expect a response sometime today?

I am attending the JCTVC meeting in Guangzhou, China, there is some inconvenience for me to access the internet.

[senseiam]

Win7 x64 - doesn't want to work with any file, mkv/ts/avc etc (error: Could not create DiAVC decoder!), I assume this is a known issue?

Edit: Updated to trial version, appears to work but still can't get .ts files to work. I get Mpeg not supported even though I am pretty sure these are valid .ts files.

[LigH]

senseiam - a Decoder comes second, after the Splitter filter for the container the MPEG4-AVC stream is contained in. First ensure that you have a working splitter filter for MPEG Transport Streams (like e.g. Haali Media Splitter - if you enabled MPEG-TS while installing it). Furthermore, there are TS files which do not contain MPEG4-AVC, but e.g. VC-1 or MPEG2; but DiAVC can only decode MPEG4-AVC.

[xv]

Nearly one month without any news, may I ask what happened? When can we expect a new (x64) version of DiAVC?

[Emulgator]

schweinsz, my DiAVC decoder 1.2.0 trial has expired as I announced some weeks ago.
I can not do any frameserving work with AVC anymore. I have paid for a license.
I would like to get a recent, working and not limited version of DiAVC, please.

P.S. OK. I see, your website now has version 1.2.1 on its download page, but still only as trial.
A non-limited version please.

[lych_necross]

schweinsz, my DiAVC decoder 1.2.0 trial has expired as I announced some weeks ago.
I can not do any frameserving work with AVC anymore. I have paid for a license.
I would like to get a recent, working and not limited version of DiAVC, please.

P.S. OK. I see, your website now has version 1.2.1 on its download page, but still only as trial.
A non-limited version please.
Seconded. I had to revert back to 1.1 once 1.2.0 expired (and I have purchased a license).

[schweinsz]

The DiAVC 1.2.1 normal version is available now.
Changes:
many bugs are fixed.
progressive lossless mode.
fix the compatibility problem with the win7/vista.

Run the diavc registration.exe, if you have registered the DiAVC, just click the button "Activate the DiAVC 1.2.1".
If you never register the DiAVC, click the button "Register the DiAVC", then click the button "Activate the DiAVC 1.2.1".
Copy the activated diavc.ax to the dgavcdecdi2001 folder, you can use the DiAVC in the dgavcdecdi2001.

I met some problems on building of the x64 version directshow filter and I try to fix it recently.
I met a compiler error when I try to add the mbaff lossless mode so that I can't finish the mbaff lossless mode.
I write a 64bit data to a buffer, but it is eatten by the ms compiler. If I print the data by printf, the decoder is ok, but if I delete the printf, it run error.

[Stephen R. Savage]

Preliminary results for lossless mode on a 712x476 source (average fps quoted):


DiAVC 1.2.1: 139 fps (112%)
CoreAVC 2.0: 125 fps (100%)
ffdshow r3507: 122 fps (97.6%)


The stream decodes correctly, and DiAVC retains a healthy advantage over its competitors. Here's to hoping that interlaced lossless and 64-bit support are coming next.

[Episodio1]

Thanks, schweinz. I was waiting for full version as well.

Thanks, Stephen, for your benchmarks. :)

[Stephen R. Savage]

@schweinsz: Reporting new bugs in DiAVC 1.2.1.

1. Playback of H.264 in AVI is just plain broken. The previous DiAVC 1.1 had no problems with this content. I have uploaded a sample here (http://www.multiupload.com/EQJ1T9JXLM). Please note how the playback speed, SAR, and cropping are not respected.

2. When seeking in MKV, DiAVC can resume with corrupted output. I previously reported this, but it happens especially frequently in the new version.

While I understand that #2 may be difficult to track down, please fix #1 ASAP.

Additionally, I have a question concerning the new "Activate DiAVC" button in "DiAVC Registration.exe". Does DiAVC now require Internet activation to be used?

[schweinsz]

@schweinsz: Reporting new bugs in DiAVC 1.2.1.

1. Playback of H.264 in AVI is just plain broken. The previous DiAVC 1.1 had no problems with this content. I have uploaded a sample here (http://www.multiupload.com/EQJ1T9JXLM). Please note how the playback speed, SAR, and cropping are not respected.
I can't connect to the site now, could you upload it to a new one?

Additionally, I have a question concerning the new "Activate DiAVC" button in "DiAVC Registration.exe". Does DiAVC now require Internet activation to be used?
No, the button "Activate the DiAVC" just modify some bytes of the diavc.ax and diavc_mmx.ax.

[Dark Shikari]

Nice to see DiAVC supporting x264 lossless now! Any reason interlaced isn't supported? As far as I can see, if you have interlaced mode, and you have progressive lossless, interlaced lossless should naturally work without any extra work unless something weird's going on.

Not like many people use interlaced lossless, I guess.

Preliminary results for lossless mode on a 712x476 source (average fps quoted):

DiAVC 1.2.1: 139 fps (112%)
CoreAVC 2.0: 125 fps (100%)
ffdshow r3507: 122 fps (97.6%)

The stream decodes correctly, and DiAVC retains a healthy advantage over its competitors. Here's to hoping that interlaced lossless and 64-bit support are coming next.

Interesting; since lossless is nearly 100% CABAC decoding, this suggests that DiAVC is about 12% faster at CABAC decoding than CoreAVC.

[schweinsz]

Nice to see DiAVC supporting x264 lossless now! Any reason interlaced isn't supported? As far as I can see, if you have interlaced mode, and you have progressive lossless, interlaced lossless should naturally work without any extra work unless something weird's going on.
Not like many people use interlaced lossless, I guess.
I finished the field lossless mode but I never tested it.
The problem is with the mbaff mode. I met a compiler error. I write a 64bit data to a buffer, ((uint64*)midbuffer)[64] = 0x3444509409302303; But I found the compiler cancel it. If I add a printf printf("%I64x\n", ((uint64*)midbuffer)[64]), the program works corectly.

[Dark Shikari]

I finished the field lossless mode but I never tested it.
The problem is with the mbaff mode. I met a compiler error. I write a 64bit data to a buffer, ((uint64*)midbuffer)[64] = 0x3444509409302303; But I found the compiler cancel it. If I add a printf printf("%I64x\n", ((uint64*)midbuffer)[64]), the program works corectly.You need to use ULL for large constants, e.g. 0x3444509409302303ULL.

[schweinsz]

You need to use ULL for large constants, e.g. 0x3444509409302303ULL.

This is not the reason.
void dec_ip_mbaff()
{
....
((uint64*)midbuffer)[64] = 0x3444509409302303;
....
printf("fst: %I64x\n", ((uint64*)midbuffer)[64]);
....
idct_stub(....);
....
}
idct_stub(....)
{
....
printf("scd: %I64x\n", ((uint64*)midbuffer)[64]);
....
}

If I don't disable the printf("fst: %I64x\n", ((uint64*)midbuffer)[64]), both printf output 0x3444509409302303, but if I disable the printf("fst: %I64x\n", ((uint64*)midbuffer)[64]), the second printf output 0.

[Stephen R. Savage]

I can't connect to the site now, could you upload it to a new one?


No, the button "Activate the DiAVC" just modify some bytes of the diavc.ax and diavc_mmx.ax.

Try one of these:

http://www.megaupload.com/?d=01Q2BUSS
http://rapidshare.com/#!download|77dt|430476302|120fps_anamorphic_AVC_AAC_SSA.avi|4645
http://depositfiles.com/en/files/udgxdm0n1
http://hotfile.com/dl/82198351/92d5fca/120fps_anamorphic_AVC_AAC_SSA.avi.html
http://www.zshare.net/video/82650806a4b53b14/
http://uploading.com/files/get/ae2e9bfc/

[schweinsz]

Try one of these:
Thank you very much. The uploading.com and the hotfile.com work.

[Dark Shikari]

If I don't disable the printf("fst: %I64x\n", ((uint64*)midbuffer)[64]), both printf output 0x3444509409302303, but if I disable the printf("fst: %I64x\n", ((uint64*)midbuffer)[64]), the second printf output 0.Oh, you mean miscompilation, not a compiler error. This is because you are violating the C rules on strict aliasing in your code, and thus your compiler has every right to cause demons to shoot out of your nose.

Having strict-aliasing violations in code is extremely dangerous.

[sherpya]

Oh, you mean miscompilation, not a compiler error. This is because you are violating the C rules on strict aliasing in your code, and thus your compiler has every right to cause demons to shoot out of your nose.

Having strict-aliasing violations in code is extremely dangerous.

this only if the buffer is not aligned to 8 byte, perhaps as far I known it's only a problem with newer gcc.
Anyway if you allocate the array as uint64 array you can still use it safely as byte buffer.

I think msvc does not obey strict aliasing rules.
You should look at the asm generated code.

[Dark Shikari]

this only if the buffer is not aligned to 8 byteNot true: on x86 alignment doesn't matter for type-puns, because unaligned loads and stores are allowed.

perhaps as far I known it's only a problem with newer gcc.Betting that a compiler doesn't abide by its own spec seems like a bad idea... I'm pretty sure gcc is not the only one that cares, considering that completely ignoring strict aliasing hurts performance by at least ~3-5%.

[sherpya]

Not true: on x86 alignment doesn't matter for type-puns, because unaligned loads and stores are allowed.

I thing we were talking about x86_64, I'm wrong?


Betting that a compiler doesn't abide by its own spec seems like a bad idea... I'm pretty sure gcc is not the only one that cares, considering that completely ignoring strict aliasing hurts performance by at least ~3-5%.

it's a problem _with_ not _of_ newer gcc
I said nothing about gcc choice, there were already enough flames I think :)

[Dark Shikari]

I thing we were talking about x86_64, I'm wrong?There's no difference; all types smaller than 128-bit do not require alignment on x86, 32 or 64.

[Kazuya]

Hello !
I'm not sure to understand, is the diAVC 1.2.1 trial version a lossless x264 decoder ?
(diavc_sse2.ax)

[zmaster]

Kazuya,
This H264 decoder. AndNote that activation is needed to use the DiAVC1.2.1.

[schweinsz]

Hello !
I'm not sure to understand, is the diAVC 1.2.1 trial version a lossless x264 decoder ?
(diavc_sse2.ax)
The DiAVC supports H.264 progressive lossless mode.

[Kazuya]

Note that activation is needed to use the DiAVC1.2.1.

Ok, then which version is the diavc_sse2.ax provided by the diavc 1.2.1_trial_version (Expires on Feb. 31, 2011) ? :confused:

[schweinsz]

Ok, then which version is the diavc_sse2.ax provided by the diavc 1.2.1_trial_version (Expires on Feb. 31, 2011) ? :confused:
No, the trial version doesn't need activation.

[acki]

What do I have to do to make the filter work ? I'm trying to register it (via menu) and it doesn't work at all. I also tried to connect it in graphstudio but refuses to accept connections....

I'm using the last trial version from the website.

Any hints ?

Thanks.

[schweinsz]

What do I have to do to make the filter work ? I'm trying to register it (via menu) and it doesn't work at all. I also tried to connect it in graphstudio but refuses to accept connections....

I'm using the last trial version from the website.

Any hints ?

Thanks.
How do you register the filter by menu? I register it by run "regsvr32.exe diavc.ax" in cmd.exe.
I unregister the diavc.ax and register it by runing the "regsvr32.exe diavc.ax", and it succeed just now.

[acki]

How do you register the filter by menu? I register it by run "regsvr32.exe diavc.ax" in cmd.exe.
I unregister the diavc.ax and register it by runing the "regsvr32.exe diavc.ax", and it succeed just now.

Well... considering it creates an entry in the main menu of winxp like this:

start menu->D¡AVC H264 Decoder->D¡AVC Registration.exe

That's what I use. Then a window pops up with 3 buttons and a textbox. Register the D¡AVC| Activate | (OK | Input password).
Then I click on Register and nothing happens.

Now I have tried regsvr32 diavc.ax with the same results: It apparently registers but refuses to work :D

I have disabled all H264 filters I have (ffdshow, DIVX h264, Coreavc) and I can't see your filter trying to decode. Again, same results in graphedit. The filter doesn't accept connections in its input pin... I'm sure I'm doing something wrong here. But what ? :confused:

Thanks for your quick answer.

[schweinsz]

Well... considering it creates an entry in the main menu of winxp like this:

start menu->D¡AVC H264 Decoder->D¡AVC Registration.exe

That's what I use. Then a window pops up with 3 buttons and a textbox. Register the D¡AVC| Activate | (OK | Input password).
Then I click on Register and nothing happens.

Now I have tried regsvr32 diavc.ax with the same results: It apparently registers but refuses to work :D

I have disabled all H264 filters I have (ffdshow, DIVX h264, Coreavc) and I can't see your filter trying to decode. Again, same results in graphedit. The filter doesn't accept connections in its input pin... I'm sure I'm doing something wrong here. But what ? :confused:

Thanks for your quick answer.

The DiAVC trial version (http://www.di-avc.com/DiAVC1.2.1_trialversion.zip) doesn't need the DiAVC Registration.exe. Just download it and run the "regsvr32.exe diavc.ax" in cmd.exe in the same directory with the diavc.ax.

[acki]

The DiAVC trial version (http://www.di-avc.com/DiAVC1.2.1_trialversion.zip) doesn't need the DiAVC Registration.exe. Just download it and run the "regsvr32.exe diavc.ax" in cmd.exe in the same directory with the diavc.ax.

Well... I tried registering manually (regsvr32 diavc.ax) and it doesn't work either. I unregistered it and re-registered it with the same results. I disabled in media player classic HT ffdshow, divx h264 and coreavc and included diavc as preferred, but nothing. Again, I tried to make a graph in graphedit and neither input or output pins connect to diavc.

It's kinda weird....

[UPDATE]
I deinstalled diavc, cleaned the registry with ccleaner, rebooted the computer, reinstalled diavc, MANUALLY registered diavc....

And now I get a messagebox saying that diavc is not activated everytime I try to use it, and of course, it does not work :devil:

Any hints ?

[Kazuya]

No, the trial version doesn't need activation.

Ok, thanks ! :)
DiAVC seems pretty good and my CPU cores work at the same level, better than CoreAVC.
I will try it longer... ;)

[audyovydeo]

Hello

I am a paying user of DiAVC 1.1.1 and neurons2's DGAVCDECDI.

I've been using both since september.
I've just downloaded, installed and re-registered 1.2.1, and DGAVCDECDI now reports :

"DiAVC Unactivated".

Note that I have DG tools (including the DiAVC.ax that ships with them) in :

\bin\DG\

and DiAVC in

\program files\DiAVC.


Can you please clarify which version should be used with DGAVCDECDI, and whether the registration process for your .ax file is system-wide or not ?

there seems to be a bit of a mess with registration.

thanks in advance
audyovydeo

[acki]

[UPDATE]
I deinstalled diavc, cleaned the registry with ccleaner, rebooted the computer, reinstalled diavc, MANUALLY registered diavc....

And now I get a messagebox saying that diavc is not activated everytime I try to use it, and of course, it does not work :devil:

Any hints ?

I forgot to mention a new "feature", If I try to insert the filter into graphstudio (after I get the pop up saying "diavc is not activated"), and connect the video output pin of matroska splitter to diavc, it FREEZES graphstudio.

I don't know if running what you add the the start menu (DiAVC registration.exe) screwed everything up... but if it did, you've obviusly lost a potential customer here since it seems I won't be able to test the trial ever. Unless you provide me with a FIX of whatever diavc did in my registry (blind guess) to avoid me using it.

[schweinsz]

The DiAVC 1.2.2 (http://di-avc.com/DiAVC1.2.2_Setup.zip) is available now.
Changes:
Fixed the bug in the DiAVC registration.

I found a problem that the machine code generated by the DiAVC Registration perhaps changes with time. Now I changes the storage address of the machine code in the registry that is different from the DiAVC filter information.
If your machine code changes, send it to me at sales@di-avc.com, I will reply to you the password to you once I receive the email.
The activation method is as follows:
Run the DiAVC Registration.exe in the same folder with the diavc.ax and the diavc_mmx.ax, fill up the password and press the button "Activate the DiAVC 1.2.1", the DiAVC Registration.exe will modify some bytes of the diavc.ax and the diavc_mmx.ax. If you find that the last modify times of the diavc.ax and the diavc_mmx.ax are changed, you can get the diavc.ax and the diavc_mmx.ax are activated. Register it then you can playback H.264 video with the DiAVC. Copy the activated diavc.ax to the DGAVCDecDI folder , you can use the DiAVC in the DGAVCDecDI.

The DiAVC trial version 1.2.2 (http://di-avc.com/DiAVC1.2.2_trialversion.zip) is available now. Hope it is OK.

[schweinsz]

I forgot to mention a new "feature", If I try to insert the filter into graphstudio (after I get the pop up saying "diavc is not activated"), and connect the video output pin of matroska splitter to diavc, it FREEZES graphstudio.
The DiAVC trial version will never pop up a dialog "diavc is not activated". You are using the unactivated DiAVC normal version. Find all diavc.ax and diavc_mmx.ax and diavc_sse2.ax in your computer and delete them, then download the DiAVC trial version 1.2.2 (http://di-avc.com/DiAVC1.2.2_trialversion.zip), register it by running "regsvr32.exe diavc.ax" in the same folder with the unzipped DiAVC trial version. Good luck for you.

[schweinsz]

Hello

I am a paying user of DiAVC 1.1.1 and neurons2's DGAVCDECDI.

I've been using both since september.
I've just downloaded, installed and re-registered 1.2.1, and DGAVCDECDI now reports :

"DiAVC Unactivated".

Note that I have DG tools (including the DiAVC.ax that ships with them) in :

\bin\DG\

and DiAVC in

\program files\DiAVC.


Can you please clarify which version should be used with DGAVCDECDI, and whether the registration process for your .ax file is system-wide or not ?

there seems to be a bit of a mess with registration.

thanks in advance
audyovydeo

Follow my directives in the link (http://forum.doom9.org/showthread.php?p=1460396#post1460396).
If you find that the machine code is changed, send it to me by e-mail, I will reply you the password.

[Stephen R. Savage]

Just a warning: If you upgrade to DiAVC 1.2.2, you have to re-enter your password, which it turns out I forgot to keep a copy of -_-.

@schweinsz: could you please send me a copy of my password via email?

Edit: Thanks, schweinsz.

[yesgrey]

The DiAVC 1.2.2 is available now.
press the button "Activate the DiAVC 1.2.1"
If DiAVC is version 1.2.2, why DiAVC Registration.exe still shows 1.2.1? Why not updating the version number on it too?:confused:

[LigH]

Probably just an ID10T error... ;)

To post a positive example between all that crying for help:

Uninstalled 1.2.0, installed 1.2.2, re-registered and re-activated DiAVC, copied AX files -- DGAVCDecDI still works. :)

[mariner]

The DiAVC trial version will never pop up a dialog "diavc is not activated". You are using the unactivated DiAVC normal version. Find all diavc.ax and diavc_mmx.ax and diavc_sse2.ax in your computer and delete them, then download the DiAVC trial version 1.2.2 (http://di-avc.com/DiAVC1.2.2_trialversion.zip), register it by running "regsvr32.exe diavc.ax" in the same folder with the unzipped DiAVC trial version. Good luck for you.

Greetings schweinsz. This is impressive work.

Tested diavc_mmc.ax in a Q8400/W7x32 configuration using a 1920x1080/60p 28mbps Panasonic clip (http://www.megaupload.com/?d=3EBAWEWN). CPU usage is practically on par with CoerAVC, just a few percentage points higher.

Does not work well with Cyberlink PDVD10 Demux 2.0.

Keep up the good work.

Best regards.

[schweinsz]

If DiAVC is version 1.2.2, why DiAVC Registration.exe still shows 1.2.1? Why not updating the version number on it too?:confused:

Ok, I updated the DiAVC Registration.exe today.

[schweinsz]

Greetings schweinsz. This is impressive work.

Tested diavc_mmc.ax in a Q8400/W7x32 configuration using a 1920x1080/60p 28mbps Panasonic clip (http://www.megaupload.com/?d=3EBAWEWN). CPU usage is practically on par with CoerAVC, just a few percentage points higher.

Does not work well with Cyberlink PDVD10 Demux 2.0.

Keep up the good work.

Best regards.

The diavc_mmx.ax is slower than the diavc.ax.

[mariner]

The diavc_mmx.ax is slower than the diavc.ax.

It is indeed. Let's see if diavc.ax is fast enough to work in an old 2.0Ghz C2D T7300 laptop.

Can you do something about crashing mpc when cyberlink demux is used?

Many thanks.

[Stephen R. Savage]

@schweinsz: Today, I tried to play a video with DiAVC 1.2.2 and I found that my machine code had changed (again) and DiAVC no longer accepts any pins in GraphStudio. Please stop fiddling with the activation code, this is getting ridiculous.

Edit: I have sent you an email with more information about this. Please sort this out ASAP, as any software that unexpectedly changes its license state is unacceptable.

[LigH]

Well, it was changed just to fix unexpected changes...

For me it did not change, it still worked. So it might be related to quite specific hardware involved in the calculation. You had bad luck, nothing worse.

[audyovydeo]

Please stop fiddling with the activation code, this is getting ridiculous.
...
any software that unexpectedly changes its license state is unacceptable.


I have to second that.
Payware has GOT to work for those who pay. Full stop.

I'm also still waiting for a reply to my question : which diavc.ax version should be used with DGAVCdecDI : the latest, or the version provided ?
If the latter, how does registration of latest version affect neuron2's (older) version ?

thanks in advance for your answers
audyovydeo

[TheRyuu]

I have to second that.
Payware has GOT to work for those who pay. Full stop.

I'm also still waiting for a reply to my question : which diavc.ax version should be used with DGAVCdecDI : the latest, or the version provided ?
If the latter, how does registration of latest version affect neuron2's (older) version ?

thanks in advance for your answers
audyovydeo

You should be able to just copy paste over, it doesn't affect dgavcdecdi at all.

(you can try it yourself!)

[LigH]

After I re-registered and re-activated DiAVC 1.2.2 (which patches both *.ax files, they will be created new on your harddisk) I copied them into the DGAVCDecDI directory -- and they worked well.

[schweinsz]

@schweinsz: Today, I tried to play a video with DiAVC 1.2.2 and I found that my machine code had changed (again) and DiAVC no longer accepts any pins in GraphStudio. Please stop fiddling with the activation code, this is getting ridiculous.

Edit: I have sent you an email with more information about this. Please sort this out ASAP, as any software that unexpectedly changes its license state is unacceptable.
The algorithm that generate the machine code is never changed. The module firstly read the allready-generated machine code from registry, if it is unavailable, the module will generate a new one (I found that it is volatile). The version before 1.2.2 store the machine code with the same position as the DiAVC filter information. The DiAVC 1.2.2 change it to another position. This is because that the regsvr32.exe can delete the DiAVC filter information with the machine code and the password once it is re-registered.
I believe the DiAVC Registration v1.2.2 will stick a exclusive machine code unless the registration information stored in registry is deleted by some other software.

[schweinsz]

I'm also still waiting for a reply to my question : which diavc.ax version should be used with DGAVCdecDI : the latest, or the version provided ?
If the latter, how does registration of latest version affect neuron2's (older) version ?

thanks in advance for your answers
audyovydeo
Use the latest version 1.2.2 for playback and DGAVCdecDI. The new version is full compatible with the DGAVCdecDI. Register it with the DiAVC registration 1.2.2 and copy the diavc.ax with the same directory as the DiAVC registration 1.2.2 to the same directory with the DGAVCdecDI.

[mariner]

Greetings schweinsz.

Tested version 1.2.2 in a C2Dt7300 2.0Ghz/GT8600M/w7x86 configuration using 28mbps 1920x1080/60p Panasonic mts clip.

Crashed MPC, switching splitter to no avail. Ran the test in Potplayer using EVR.

Results are impressive. CPU usage hovers at around 100%. Is there any room for a further 5-10% improvement?

Best regards.

[schweinsz]

Greetings schweinsz.

Tested version 1.2.2 in a C2Dt7300 2.0Ghz/GT8600M/w7x86 configuration using 28mbps 1920x1080/60p Panasonic mts clip.

Crashed MPC, switching splitter to no avail. Ran the test in Potplayer using EVR.

Results are impressive. CPU usage hovers at around 100%. Is there any room for a further 5-10% improvement?

Best regards.

Perhaps switch to intel c/c++ compiler and turn on the profile based optimization could help some.
The vc2010 is super bad. The new build is without profile based optimization.

[mariner]

Perhaps switch to intel c/c++ compiler and turn on the profile based optimization could help some.
The vc2010 is super bad. The new build is without profile based optimization.

Looking forward to the new build.

Could not resolve the weird MPC crashing problem. Stability issue, perhaps?

Many thanks and best regards.

[Daemon404]

To anyone wondering about why DiAVC's machine code seems to randomly change, here's a nice story:

When I generated the machine code I used to register DiAVC, I had an external harddrive plugged in, and turned on. Later I noticed my machine code had changed, and that turning that external drive on again made it go back. So, it uses info from current drives and whatnot. This is terribly restrictive and it only furthers the hate towards BS hardware-tied crap. (And it should.) It's also not as if obtaining a new serial is quick.. it has to be done manually by the author.

I for one regret paying for such a restrictive software. Hardware-tied crap needs to die.

[schweinsz]

To anyone wondering about why DiAVC's machine code seems to randomly change, here's a nice story:

When I generated the machine code I used to register DiAVC, I had an external harddrive plugged in, and turned on. Later I noticed my machine code had changed, and that turning that external drive on again made it go back. So, it uses info from current drives and whatnot. This is terribly restrictive and it only furthers the hate towards BS hardware-tied crap. (And it should.) It's also not as if obtaining a new serial is quick.. it has to be done manually by the author.

I for one regret paying for such a restrictive software. Hardware-tied crap needs to die.
The current version v1.2.2 is partly hardware-tied. The DiAVC Registration read a harddisk serial number and store it in the registry. The machine code and password is with the same position as the DiAVC filter information before the v1.2.1 so that when the filter is re-registered, the machine code information will be deleted. Now I change it to different position. If there is the machine code in the registry, the DiAVC Registration will read the machine code directly from the registry and will not generate a new one (while the later can give different machine code for its every run).
The DiAVC Registration change some bytes when activating the DiAVC filter. The DiAVC filter doesn't depend on the harddisk.
At last, I will change the registration in the next version so that it entirely doesn't on any hardware.

[sneaker_ger]

Good to hear. This DRM is a PITA.

Did you look into 10 bit decoding yet? x264 has had support for a while now, but not a single decoder out there supports it yet.

[kieranrk]

Good to hear. This DRM is a PITA.

Did you look into 10 bit decoding yet? x264 has had support for a while now, but not a single decoder out there supports it yet.

Mainconcept broadcast does.

[schweinsz]

Good to hear. This DRM is a PITA.

Did you look into 10 bit decoding yet? x264 has had support for a while now, but not a single decoder out there supports it yet.

No plan to support 10bit in the next one year. It is very complex.

[madshi]

I think the Divx decoder decodes 10bit just fine, but rounds it down to 8bit. Not 100% sure, though.

[prokhozhijj]

I am using DiAVCDEcode (by Donald Graft, aka neuron2) and DiAVC decoder (trial version). I've dowloaded both products recently so both products have last versions.

There is my Avisynth script


SetMemoryMax(1200)
SetMTMode(5,4)

DGSource("my_file.dgi")
#DSS2("my_file.mkv")

SetMTMode(2)

super = MSuper(pel=2, sharp=2)
backward_vec2 = MAnalyse(super, isb=true, delta=2, overlap=4)
backward_vec1 = MAnalyse(super, isb=true, delta=1, overlap=4)
forward_vec1 = MAnalyse(super, isb=false, delta=1, overlap=4)
forward_vec2 = MAnalyse(super, isb=false, delta=2, overlap=4)
MDegrain2(super, backward_vec1,forward_vec1,backward_vec2,forward_vec2,thSAD=400)
return last


It crashes all time when I try to process my video, in the very beginning, just before encoding. If multi-threading turned off than script runs smoothly but with one core only in use (I have quad-core CPU).
Meanwhile, if I replace DGSource with DSS2 (it comes with HAALI media splitter, avss.dll) script runs smoothly in multi-threaded mode too with full processor loading.

Donald told me that he is not interested in investigation what is wrong with MT plug-in. He knows nothing about it. There is our dialog with Donald. http://neuron2.net/board/viewtopic.php?f=6&t=70&start=10

Is it possible to turn on multi-threading for my script with DiAVC and DIAVCDecode?

[Stephen R. Savage]

@prokhozhijj: If Avisynth MT works with DSS2 but not DGSource, the problem is that Graft's code is not thread-safe. Fixing it would require code modifications to DGSource. One workaround you could try if you have sufficient memory would be splitting your script into four parts with the Trim() function and starting a new encoder process for each part.

[prokhozhijj]

One workaround you could try if you have sufficient memory would be splitting your script into four parts with the Trim() function and starting a new encoder process for each part.

Thank you. It looks like this is only solution.


Is there any way how to force to use multithreading in trial version of DiAVC? Neuron2 said that it is possible to turn on multithreading in DiAVC decoder. But I have no idea how to do it for trial version. May be some settings in registry?

[Stephen R. Savage]

DiAVC is internally multi-threaded, but that has nothing to do with Avisynth MT, which creates multiple instances of filters (DGSource). There is a registry key that sets all DiAVC options, but it is encoded with some kind of bitmask, so it may be difficult to change the thread count by itself. However, the default behavior is to open 1.5 decoder threads per CPU (I think), and you do not need to change it. Your problem is with DGSource and its incompatibility with Avisynth MT. Besides, in the script that you posted, the vast majority of CPU time is spent in MVTools.

P.S. By all reports, calling SetMTMode(5) before DGSource is supposed to fix compatibility issues. Have you tried it?

[prokhozhijj]

P.S. By all reports, calling SetMTMode(5) before DGSource is supposed to fix compatibility issues. Have you tried it?

Of course I've tried. Just look on my previous post. It crashes. Even with new version that Donald just made.

[WILLIS]

are their any alternative gui interfaces that work alongside di-avc other than dg's decoder? does di-avc come with a native gui?

[LoRd_MuldeR]

are their any alternative gui interfaces that work alongside di-avc other than dg's decoder? does di-avc come with a native gui?

What kind of GUI do you need? DiAVC is implemented as a DirectShow Filter. It can (should) work with any DirectShow-based Player software you like, for example MPC-HC.

The purpose of DGAVCDecDI is making the DiAVC decoder available as a source filter in Avisynth - through it's native interface, rather than the DirectShow interface.

I'm not aware of any other software that is built on top of DiAVC's native interface. However there is an SDK available on the DiAVC web-site, so everybody could create such software.

But it has to be noted that for any DiAVC-based software each user would need to obtain a license of DiAVC first, as is the case with DGAVCDecDI...

[GREG1292]

Sounds stupid but what does this codec in real world terms
gain me over CoreAvc when using frame doubling script,madvr.

How much will this lesson my cpu load. Don't know all the questions to ask but am just curious before I try it out so
I know what to expect?

Thanks Greg

[schweinsz]

Sounds stupid but what does this codec in real world terms
gain me over CoreAvc when using frame doubling script,madvr.

How much will this lesson my cpu load. Don't know all the questions to ask but am just curious before I try it out so
I know what to expect?

Thanks Greg
The gain depends on your CPU. You can refer to my comparison (http://di-avc.com/compare/compare.html).
When the CPU is Core 2 Duo, the gain is about 20%.
If the CPU is old, the gain is small.

[GREG1292]

The gain depends on your CPU. You can refer to my comparison (http://di-avc.com/compare/compare.html).
When the CPU is Core 2 Duo, the gain is about 20%.
If the CPU is old, the gain is small.

I use an amd 1055t hex core which is newer and with
CoreAvc and frame doubling on my preferred script 88% cpu usuage. Trying
to figure if it will save me enough headroom on my
cpu or buy a i7

[Episodio1]

GREG, you can download trial version and check for yourself.

[lych_necross]

I use an amd 1055t hex core which is newer and with
CoreAvc and frame doubling on my preferred script 88% cpu usuage. Trying
to figure if it will save me enough headroom on my
cpu or buy a i7

Cpu usage is not an accurate measure of performance. Try comparing the number of frames decoded per second between coreavc and diavc for an accurate measure of performance.

[schweinsz]

I use an amd 1055t hex core which is newer and with
CoreAvc and frame doubling on my preferred script 88% cpu usuage. Trying
to figure if it will save me enough headroom on my
cpu or buy a i7

You can download the DiAVC trial version (http://www.di-avc.com/DiAVC1.2.3_trialversion.zip), and register it by running "regsvr32.exe diavc.ax", then you can use it.

[yesgrey]

the trial version is numbered 1.2.3 but the regular version is numbered 1.2.2. Is the trial newer? What has changed?

[schweinsz]

the trial version is numbered 1.2.3 but the regular version is numbered 1.2.2. Is the trial newer? What has changed?
I added a small feature named "low delay". If you playback video generally on local machine, don't choose it.

[Forfront]

I use an amd 1055t hex core which is newer and with
CoreAvc and frame doubling on my preferred script 88% cpu usuage. Trying
to figure if it will save me enough headroom on my
cpu or buy a i7

GREG1292 ~ I have seen ur name before on the "frame doubling" thread on avsforum, do u still have a GTX 460??

[mariner]

MBAFF Interlaced video

Greetings schweinsz.

DiAVC seems to have problems dealing with MBAFF interlaced video such as this Beyonce clip (http://mirror05.x264.nl/public/beyonce.at.the.bbc.1080mbaff.sample.ts).

http://forum.doom9.org/showpost.php?p=927671&postcount=3

[schweinsz]

MBAFF Interlaced video

Greetings schweinsz.

DiAVC seems to have problems dealing with MBAFF interlaced video such as this Beyonce clip (http://mirror05.x264.nl/public/beyonce.at.the.bbc.1080mbaff.sample.ts).

http://forum.doom9.org/showpost.php?p=927671&postcount=3

What did you see? I play it and find the DiAVC deal with it fine.

[mariner]

It plays with no de-interlacing at 29.97 fps, like weave.

[Stephen R. Savage]

@GREG1292: CoreAVC will probably have lower total CPU usage due to PureVideo.

@mariner: DiAVC does no deinterlacing on its own. In an earlier post, schweinsz mentioned that it just sets DEINTERLACE_BOB for handling downstream.

[mariner]

@mariner: DiAVC does no deinterlacing on its own. In an earlier post, schweinsz mentioned that it just sets DEINTERLACE_BOB for handling downstream.

Thanks for the reply, Stephen.

What do you think caused the problem?

[LigH]

Where is the problem?

DiAVC decodes. Correctly. That's its only job.

If you want a deinterlaced output, use a deinterlacer in addition to the decoder.

One possible solution for some Media Player could be ffdshow if you enable it for raw video and set it up to deinterlace automatically for specific resolutions.

[LoRd_MuldeR]

One possible solution for some Media Player could be ffdshow if you enable it for raw video and set it up to deinterlace automatically for specific resolutions.

...or simply use a renderer that will honor the DEINTERLACE_BOB flag and deinterlace.

[kieranrk]

...or simply use a renderer that will honor the DEINTERLACE_BOB flag and deinterlace.

And enjoy your poor gpu deinterlacing quality...

[rajs]

Schweinsz,

Would you mind taking a look at the following sample?

http://www.megaupload.com/?d=SYAZK4QQ

When playing back in DiAVC, there are blocking artifacts throughout the 2nd GOP. Other H.264 decoders (e.g. ffmpeg-mt) play back this sample OK.

The sample was produced by a H.264 editing program which re-codes only those frames near the cut point that need to be re-encoded. In this sample, the first GOP will have been re-encoded but probably not the second.

I'd be very interested to know if this sample is violating H.264 in any way.

:thanks:

[schweinsz]

Schweinsz,

Would you mind taking a look at the following sample?

http://www.megaupload.com/?d=SYAZK4QQ

When playing back in DiAVC, there are blocking artifacts throughout the 2nd GOP. Other H.264 decoders (e.g. ffmpeg-mt) play back this sample OK.

The sample was produced by a H.264 editing program which re-codes only those frames near the cut point that need to be re-encoded. In this sample, the first GOP will have been re-encoded but probably not the second.

I'd be very interested to know if this sample is violating H.264 in any way.

:thanks:
I believe that you provide a corrupted bitstream. The JM can't decode it and it crashes. The DiAVC can decode it but with artifacts.

[rajs]

Thanks.

[tormento]

Any ETA?

[schweinsz]

Any ETA?
I plan to finish it before the end of this June. I am working on a project for a company and previously plan to finish it in this February but I finish it now.

[schweinsz]

I finished the feature "Dynamic Format Change".
Now the DiAVC trial version 1.2.5 (http://www.di-avc.com/DiAVC1.2.5_trialversion.7z) that includes the feature "Dynamic Format Change" is available in di-avc.com.
If everything is OK, I will release the DiAVC retail version 1.2.5.

In addition, I will release the x64 version at the end of this June.

[mariner]

I finished the feature "Dynamic Format Change".
Now the DiAVC trial version 1.2.5 (http://www.di-avc.com/DiAVC1.2.5_trialversion.7z) that includes the feature "Dynamic Format Change" is available in di-avc.com.
If everything is OK, I will release the DiAVC retail version 1.2.5.

In addition, I will release the x64 version at the end of this June.

Thanks for the new version, schweinsz.

The MBAFF problem has not been solved. See this:

http://img192.imageshack.us/i/beyonceue.jpg/
http://yfrog.com/5cbeyonceuej


Best regards.

[Selur]

In addition, I will release the x64 version at the end of this June.
Nice :)

[Episodio1]

What does "Dynamic Format Change" do ?

[LigH]

I could guess it is related to switching video stream attributes (resolution, aspect ratio) like in transport streams of TV broadcasts with inserted advertizing during a movie?

[schweinsz]

What does "Dynamic Format Change" do ?

The video's resolution or SAR changes along the time.

[06_taro]

When using DiAVC 1.2.2 with latest madvr(0.61) & MPC-HC(1.5.2.3149, xvidvideo build), the internal subtitles filter of mpc-hc doesn't work.

There's no problem when using DiAVC with other renderer like Haali's Renderer or EVR-CP, or using other AVC decoder with madvr instead of DiAVC. And of course I can use directvobsub instead of internal filter, but sometimes I need the keyboard shortcut to adjust the delay time of subtitles.

Is it an issue of recent DiAVC or just caused by my wrong setting? I know the madvr's update of using mpc-hc's internal filter was much later than the release date of DiAVC 1.2.2, but I haven't test with DiAVC 1.2.5 trial version....

[schweinsz]

When using DiAVC 1.2.2 with latest madvr(0.61) & MPC-HC(1.5.2.3149, xvidvideo build), the internal subtitles filter of mpc-hc doesn't work.

There's no problem when using DiAVC with other renderer like Haali's Renderer or EVR-CP, or using other AVC decoder with madvr instead of DiAVC. And of course I can use directvobsub instead of internal filter, but sometimes I need the keyboard shortcut to adjust the delay time of subtitles.

Is it an issue of recent DiAVC or just caused by my wrong setting? I know the madvr's update of using mpc-hc's internal filter was much later than the release date of DiAVC 1.2.2, but I haven't test with DiAVC 1.2.5 trial version....

Could you share me a small H.264 clip with subtitles.
I have not such bitstreams in hand.

[06_taro]

test5-secondary audio commentary track, misc subtitle tracks.rar (http://www.mediafire.com/file/i3t39ph6kpx90vn/test5-secondary%20audio%20commentary%20track%2C%20misc%20subtitle%20tracks.rar)
Here's a clip with subtitles from matroska's test suite.
DiAVC works fine with mpc-hc's internal subtitles filter and Haali's Renderer or EVR-CP, so do other AVC decoders with mpc-hc's internal filter and madVR. But I can never get the subtitles shown when using DiAVC 1.2.2 with madvr except for using DirectVobSub.

[schweinsz]

test5-secondary audio commentary track, misc subtitle tracks.rar (http://www.mediafire.com/file/i3t39ph6kpx90vn/test5-secondary%20audio%20commentary%20track%2C%20misc%20subtitle%20tracks.rar)
Here's a clip with subtitles from matroska's test suite.
DiAVC works fine with mpc-hc's internal subtitles filter and Haali's Renderer or EVR-CP, so do other AVC decoders with mpc-hc's internal filter and madVR. But I can never get the subtitles shown when using DiAVC 1.2.2 with madvr except for using DirectVobSub.
Could you re-upload it to http://www.megaupload.com/?
I found that I can't connect to the mediafire still.

[schweinsz]

Thanks for the new version, schweinsz.

The MBAFF problem has not been solved. See this:

http://img192.imageshack.us/i/beyonceue.jpg/
http://yfrog.com/5cbeyonceuej


Best regards.
I decide to code a software de-interlacing algorithm recently.

[kolak]

When I want to buy license through your website details for PayPal account are processed on non secured website- not good.

Andrew

[Underground78]

Could you re-upload it to http://www.megaupload.com/?
I found that I can't connect to the mediafire still.

http://www.megaupload.com/?d=Q4CZVG54

[06_taro]

MU：test5-secondary audio commentary track, misc subtitle tracks.rar (http://www.megaupload.com/?d=K5J13T45)
actually i'm not sure whether this issue is caused by diavc or madvr....

[schweinsz]

The DiAVC v1.2.6 (http://www.di-avc.com/DiAVC1.2.6_Setup.zip) is available on di-avc.com.
Change:
Add the support for dynamic format change.

I will try to code a software de-interlacing algorithm recently.

[madshi]

I will try to code a software de-interlacing algorithm recently.
IVTC or video mode deinterlacing or both? Is it really worth it, considering that ATI/NVidia DXVA deinterlacing works reasonably well?

[schweinsz]

IVTC or video mode deinterlacing or both? Is it really worth it, considering that ATI/NVidia DXVA deinterlacing works reasonably well?
Do you think that it is useless? Then I can concentrate on the x64 version.

[madshi]

I wouldn't say "useless".

A lot depends on whether you want to deinterlace movies or video content (or both), because *totally* different kind of deinterlacers are needed in either case.

For movies you need IVTC, and IMHO there's not a good IVTC deinterlacer available for h264 at the moment. Even the NVidia/ATI hardware IVTC isn't optimal because they are not able to reduce frame rate from 60i to 24p, AFAIK. Instead they produce 30p or 60p output, I believe, so the typical NTSC 3:2 motion judder is still there. Implementing proper 60i -> 24p IVTC in software shouldn't be too hard, so that might be something worth thinking about. You could compare the DScaler IVTC Mod decoder source code to the original DScaler decoder source code to get some ideas on how to implement IVTC inside of your decoder, it's really not very hard to do.

However, video mode deinterlacing is a whole different beast. With movies/IVTC all the pixels are there and it's only a matter of weaving the correct fields together. In contrast with video content half of the pixels are really missing and you have to find a good way to "guess"/interpolate them, which is not too easy if you want to achieve best possible quality. Also there are some relatively good video mode deinterlacers out there already, e.g. yadif (ffdshow). So I'm not sure if writing another video mode deinterlacer would be worth it, especially because Nvidia & ATI have a pretty good hardware deinterlacing implementation for video mode content.

So to sum up: Writing a quick and easy IVTC algorithm might be worth a thought, because it's relatively easy to do and would bring a good benefit. I would recommend against writing a video mode deinterlacing, though.

Another problem is this: Writing an IVTC deinterlacer and/or a video mode deinterlacer alone is one thing. If you want the user to lay back and not have to change settings all the time, you'd even have to implement an automatic movie / video content detection. You might even have to support mixed movie / video content. This would all be getting very very complicated and time consuming. If you do IVTC, only, that'd be fine and should be easy to do, but the user would have to manually switch IVTC on/off whenever he's switching between movie and video content.

Personally, I would rather like to see you work on x64, 9/10bit decoding and eventually MVC decoding, instead of working on a deinterlacer. But that's just my personal opinion. FWIW, libav/ffmpeg are supporting 9/10bit decoding now, CoreAVC has it on its to do list. So if you don't do it you'll be falling behind your competition. There isn't much native 9/10bit content out there today. But it seems that even with original 8bit content (e.g. Blu-Ray), if you re-encode it, encoding to 9/10bit seems to help compression efficiency. So I think in the future we might see more (reencoded) 9/10bit h264 content appear.

[CruNcher]

I wouldn't say "useless".

A lot depends on whether you want to deinterlace movies or video content (or both), because *totally* different kind of deinterlacers are needed in either case.

For movies you need IVTC, and IMHO there's not a good IVTC deinterlacer available for h264 at the moment. Even the NVidia/ATI hardware IVTC isn't optimal because they are not able to reduce frame rate from 60i to 24p, AFAIK. Instead they produce 30p or 60p output, I believe, so the typical NTSC 3:2 motion judder is still there. Implementing proper 60i -> 24p IVTC in software shouldn't be too hard, so that might be something worth thinking about. You could compare the DScaler IVTC Mod decoder source code to the original DScaler decoder source code to get some ideas on how to implement IVTC inside of your decoder, it's really not very hard to do.

However, video mode deinterlacing is a whole different beast. With movies/IVTC all the pixels are there and it's only a matter of weaving the correct fields together. In contrast with video content half of the pixels are really missing and you have to find a good way to "guess"/interpolate them, which is not too easy if you want to achieve best possible quality. Also there are some relatively good video mode deinterlacers out there already, e.g. yadif (ffdshow). So I'm not sure if writing another video mode deinterlacer would be worth it, especially because Nvidia & ATI have a pretty good hardware deinterlacing implementation for video mode content.

So to sum up: Writing a quick and easy IVTC algorithm might be worth a thought, because it's relatively easy to do and would bring a good benefit. I would recommend against writing a video mode deinterlacing, though.

Another problem is this: Writing an IVTC deinterlacer and/or a video mode deinterlacer alone is one thing. If you want the user to lay back and not have to change settings all the time, you'd even have to implement an automatic movie / video content detection. You might even have to support mixed movie / video content. This would all be getting very very complicated and time consuming. If you do IVTC, only, that'd be fine and should be easy to do, but the user would have to manually switch IVTC on/off whenever he's switching between movie and video content.

Personally, I would rather like to see you work on x64, 9/10bit decoding and eventually MVC decoding, instead of working on a deinterlacer. But that's just my personal opinion. FWIW, libav/ffmpeg are supporting 9/10bit decoding now, CoreAVC has it on its to do list. So if you don't do it you'll be falling behind your competition. There isn't much native 9/10bit content out there today. But it seems that even with original 8bit content (e.g. Blu-Ray), if you re-encode it, encoding to 9/10bit seems to help compression efficiency. So I think in the future we might see more (reencoded) 9/10bit h264 content appear.

This would be a heavy split and indeed it could happen like back then with DivX, user content dominating a move to 10 bit this time for consumers it will be interesting to watch if that happens, indeed it would be slowly time to leave the 8 bit age for consumer Video use and approach the 10 bit age (new devices,new software) and i mean after the 3D Buzz industry needs something new and that could be very well moving the professional 10 Bit usage into consumer space finally (not to forget a buzz marketing name for it) and even appeal user that way that didn't found the 3D move interesting @ all ;)
The biggest OSS Video projects (most influencing) already started on that move anyways :)
And yeah the time to implement a own Software Realtime Deinterlacing in times of GPUs seems wasted, and if you need deinterlacing software wise then implementing yadiff seems the most efficient choice for not wasting to much time :)

[schweinsz]

I think that I should concentrate on the x64 version.

[schweinsz]

MU：test5-secondary audio commentary track, misc subtitle tracks.rar (http://www.megaupload.com/?d=K5J13T45)
actually i'm not sure whether this issue is caused by diavc or madvr....
I tested the sample by the graphstudio. When the filter chain is haali media splitter-DiAVC or CoreAVC-renderer, both are without the subtitles. When the DirectVobSub is inserted between the renderer and the decoder, both are with subtitles. What is your filter chain when you play the sample with CoreAVC or DiAVC?

[mariner]

Thanks for the new version, schweinsz.

The MBAFF problem has not been solved. See this:

http://img192.imageshack.us/i/beyonceue.jpg/
http://yfrog.com/5cbeyonceuej

Best regards.

I decide to code a software de-interlacing algorithm recently.

Thanks for the new build, schweinsz.

Most decoders do not provide software de-interlacing functionality, relying instead on the GPU hardware to do the job by raising the appropriate interlace flag. Isn't this something relatively straight forward to add to DiAVC?

Best regards.

[ForceX]

I tested the sample by the graphstudio. When the filter chain is haali media splitter-DiAVC or CoreAVC-renderer, both are without the subtitles. When the DirectVobSub is inserted between the renderer and the decoder, both are with subtitles. What is your filter chain when you play the sample with CoreAVC or DiAVC?

I am having similar problems where subtitles do not work. Regardless of whether I use Haali splitter or MPC-HC's built-in Matroska splitter, subtitles just do not work with DiAVC+madVR, when using MPC-HC's internal subtitle renderer.

Of course, if you're using Graphstudio, you need to put DirectVobSub in the filterchain, or there would not be any subtitle renderer. But MPC-HC has its built-in subtitle renderer, so you don't need to use DirectVobSub. However, MPC-HC's subtitle renderer is not working with DiAVC+madVR, it doesn't matter which splitter is used. Any other decoder (CoreAVC/FFDshow/DivX) works with madVR and MPC-HC's internal subtitle renderer.

[schweinsz]

Most decoders do not provide software de-interlacing functionality, relying instead on the GPU hardware to do the job by raising the appropriate interlace flag. Isn't this something relatively straight forward to add to DiAVC?

Best regards.
I can provide the option in the property page, it is easy.

[tormento]

I think that I should concentrate on the x64 version.
Oh yes. ;)

[mariner]

I can provide the option in the property page, it is easy.

That would be a good start, before you embark on the ambitious software de-interlacer project.

JanWillem32 has just posted something which you may find useful:

A slight correction, deinterlacing is a mixer/renderer task, not a decoding task. There are complex and less complex deinterlacing methods. The basic types hardly require any processing power at all.
For processing, the VMR-9 and EVR mixer can offload this to the GPU. Deinterlacing was one of the first GPU video filters to be implemented for usage with VMR.
From a correct rendering perspective, it's wrong to deinterlace while the chroma is still sub-sampled (4:2:0 or 4:2:2), as that will always lower chroma quality (saturation, sharpness in color contrasts, color-to-grayscale resolution and color accuracy). That's why it's quite understandable that DirectShow and Media Foundation allow interlacing flags for decoders to signal the mixer that it should perform deinterlacing. It's unfortunate that still not all decoders offer an interlacing-compatible color format and set the correct interlacing flags on the output pin (even while DirectShow exists since the introduction of Windows 98).
For even more advanced solutions, a renderer shouldn't use a non-linear colorspace (like those used with most encoded videos) when performing color blending and no 8-bit working surfaces, because of the big quantization errors. I'm currently trying to improve the current implementation for the mixing tasks in the VMR-9 r. and EVR CP renderers, so I've been studying these subjects a bit lately.

I've just downloaded an interlaced VC-1 sample, it plays fine with the default Windows 7 codec. For the output pin:Filter : WMVideo Decoder DMO - CLSID : {94297043-BD82-4DFD-B0DE-8177739C6D20}

- Connected to:

CLSID: {FA10746C-9B63-4B6C-BC49-FC300EA5F256}
Filter: Enhanced Video Renderer (custom presenter)
Pin: EVR Input0

- Connection media type:

Video: NV12 2048x1080 (16:9) 25.00fps

AM_MEDIA_TYPE:
majortype: MEDIATYPE_Video {73646976-0000-0010-8000-00AA00389B71}
subtype: MEDIASUBTYPE_NV12 {3231564E-0000-0010-8000-00AA00389B71}
formattype: FORMAT_VideoInfo2 {F72A76A0-EB0A-11D0-ACE4-0000C0CC16BA}
bFixedSizeSamples: 1
bTemporalCompression: 0
lSampleSize: 3317760
cbFormat: 1152

VIDEOINFOHEADER:
rcSource: (0,0)-(1920,1080)
rcTarget: (0,0)-(1920,1080)
dwBitRate: 0
dwBitErrorRate: 0
AvgTimePerFrame: 400000

VIDEOINFOHEADER2:
dwInterlaceFlags: 0x000000a5
dwCopyProtectFlags: 0x00000000
dwPictAspectRatioX: 16
dwPictAspectRatioY: 9
dwControlFlags: 0x00000000
dwReserved2: 0x00000000

[schweinsz]

I am sorry that the x64 version is not finished still. I am delayed by some things in the middle of the month. The x64 version will be available before July 12 in my current estimation.

[LigH]

No panic.

Quality over speed.

[tormento]

I am sorry that the x64 version is not finished still. I am delayed by some things in the middle of the month. The x64 version will be available before July 12 in my current estimation.
It's ok, pal.

[schweinsz]

The DiAVC x64 first version (http://www.di-avc.com/diavc_x64.zip) is finished. I provide a totally free version now. It is free for non-comercial and personal use. It is without any ads. and time limitation.
The current x64 version is without lossless support and DG tools support. I will add it recently.

[schweinsz]

I plan to add 9, 10bits support for the DiAVC next.

[Underground78]

I provide a totally free version now. It is free for non-comercial and personal use. It is without any ads. and time limitation.

Nice ! :) Will you release a free version for x86 ?

Also have you some benchmark for x64 vs x86 ?

[BetaBoy]

schweinsz, with respect I am confused. Are you an MPEGLA licensee for AVC? Reason I ask is the last time I checked, you can not give away a commercial based product for free (well technically you can up to 100k, but you still need to identify each unit paid for or downloaded for free against that 100k).

[schweinsz]

Nice ! :) Will you release a free version for x86 ?

Also have you some benchmark for x64 vs x86 ?

How to benchmark the x64 codec? I found that the DiAVC cmd x64 is faster than the DiAVC x86 by about 5%.

[TheRyuu]

How to benchmark the x64 codec? I found that the DiAVC cmd x64 is faster than the DiAVC x86 by about 5%.

LoadPlugin("DirectShowSource.dll")
DirectShowSource("hurfdurf.mkv")

avs2avi filename.avs -c null -o n

[Stephen R. Savage]

schweinsz, with respect I am confused. Are you an MPEGLA licensee for AVC? Reason I ask is the last time I checked, you can not give away a commercial based product for free (well technically you can up to 100k, but you still need to identify each unit paid for or downloaded for free against that 100k).

schweinsz happens to be located in the progressive country of China, and is therefore not bound by patent or copyright law. Tough luck, eh? That's how globalization works.

@schweinsz: I am unable to use the x64 DiAVC when logged in as a regular user, it only initializes when I use "Run as administrator". When I try to add the filter in GraphStudio, it refuses to connect to anything, and MPC-HC can not add it to its graph. Please check that DiAVC x64 is not attempting to do any privileged operations.

Benchmarks:

Test clip: 1920x1080p, ~10 Mbps, 8-bit AVC, no weighted P-frames
CPU: Intel Core 2 Duo T7250 "Merom" (2.00 GHz)
OS: Microsoft Windows Server 2008 R2 Standard

GraphStudio 0.3.2.0 was used for all measurements.
Reported values are the average of three independent trials.
Standard deviation was below 1% of mean for all tested software.
Output accuracy was evaluated visually for each decoder.

CoreAVC 2.5.5:
x86 - 60.4 fps
x64 - 59.4 fps

DiAVC 1.2.6:
x86 - 67.3 fps
x64 - 66.5 fps

ffdshow r3947 (XhmikosR):
x86 - 61.8 fps
x64 - 58.3 fps


I am unable to verify any performance improvements due to 64-bit decoding. This may be because of my CPU, because 64-bit DirectShow has high overhead, or simply because 64-bit is not useful for video decoding.

Edit: Never mind, it seems that DiAVC 64-bit only needs to be run as administrator for the first run. Probably because it's trying to create some registry keys in HKLM and gives up if it fails (which it shouldn't do; it should continue execution).

P.S. I look forward to 10-bit support in DiAVC, schweinsz. CoreCodec and ffmpeg have both already added support in their development versions, and with the 50%+ increase in decoding cost, this could greatly enhance the market value of a fast decoder like yours. Please work on this with all haste. While not as urgent, I hope you are also investigating adding support for the 4:4:4 colorspace.

[BetaBoy]

Stephen.... From day one, the distribution of Di-AVC has always been done by a US based web host and as such (in my opinion) falls under the responsibilities of paying for patents and royalties.

[Jarod Middelman]

schweinsz, if you plan on releasing both 32bit and 64bit version for free, i can help you host it and actively promote it.

x264.nl is ready for you!

[schweinsz]

schweinsz, if you plan on releasing both 32bit and 64bit version for free, i can help you host it and actively promote it.

x264.nl is ready for you!

Greatly! I DO plan to release a 32bit free version with reduced function. The free 64bit version is on now, You can promote it now.

[schweinsz]

P.S. I look forward to 10-bit support in DiAVC, schweinsz. CoreCodec and ffmpeg have both already added support in their development versions, and with the 50%+ increase in decoding cost, this could greatly enhance the market value of a fast decoder like yours. Please work on this with all haste. While not as urgent, I hope you are also investigating adding support for the 4:4:4 colorspace.
The DiAVC 64bits only read/write the registry on the information of the filter parameters (brightness, contrast.....).
I am working on the 9-10bits support now.
But I hestitate on the value of the 4:4:4 colorspace.

[Stephen R. Savage]

Greatly! I DO plan to release a 32bit free version with reduced function. The free 64bit version is on now, You can promote it now.

schweinsz, I am confused as to how you plan on offering a "reduced function" version of DiAVC. Currently, DiAVC does exactly one thing: decode H.264. If you released a version with intentionally crippled compatibility or performance, I fail to see how anybody could find it useful. Perhaps you could elaborate on this so that we might all understand what you are thinking?

As for the registry and the 64-bit version, try this:

1. Register DiAVC 64-bit on a machine that has never had DiAVC 64-bit installed.
2. Open GraphStudio as a non-administrator user.
3. Attempt to create a graph with DiAVC 64-bit.

DiAVC will refuse to initiate. However, if you run GraphStudio as an administrator, build a graph with DiAVC, then re-open GraphStudio as a non-administrator, it will succeed.

[upyzl]

@Stephen R. Savage
How did you registry it, please?

After successful registry http://i.imgur.com/fzSgk.png
I still can't find DiAVC... http://i.imgur.com/dzrxn.png

Of course all of OPs are under administrator

My OS is Win7 x64 SP1
----
and...
ffdshow r3947 (XhmikosR)
where did you get it? (I didn't find it on sourceforge...)

[fairchild]

I am also having problems getting the x64 filter to work with the latest mpc-hc. I registered it correctly and am able to load it as an external filter and even double click it and set it's options up. The problem is that it will only load and use the decoder if I run mpc-hc with admin rights. If I run it normally it defaults to the next decoder that is available.

Also, thanks for providing this free.

[schweinsz]

schweinsz, I am confused as to how you plan on offering a "reduced function" version of DiAVC. Currently, DiAVC does exactly one thing: decode H.264. If you released a version with intentionally crippled compatibility or performance, I fail to see how anybody could find it useful. Perhaps you could elaborate on this so that we might all understand what you are thinking?


The "reduced function" refers to the SDK support, 9-10bits decoding.

[Stephen R. Savage]

@Stephen R. Savage
How did you registry it, please?

After successful registry http://i.imgur.com/fzSgk.png
I still can't find DiAVC... http://i.imgur.com/dzrxn.png

Of course all of OPs are under administrator

My OS is Win7 x64 SP1
----
and...
ffdshow r3947 (XhmikosR)
where did you get it? (I didn't find it on sourceforge...)

Are you using 64-bit GraphStudio? It should be listed as "DiAVC 64-bit". XhmikosR builds of ffdshow and MPC-HC are available at http://xhmikosr.1f0.de/ but performance is similar to official svn pre-releases.

[hajj_3]

If you are giving away the 32bit and 64bit versions maybe they could be integrated in MPC-HC and VLC by default or something?

[Stephen R. Savage]

If you are giving away the 32bit and 64bit versions maybe they could be integrated in MPC-HC and VLC by default or something?

Well, probably he would not be willing to release the source code, as he sells SDK licenses, so that would probably be a non-starter.

@schweinsz: Can you elaborate a bit on the future 9/10-bit support? Will DiAVC output be at 8, 10, or 16-bit for these streams? If outputting at 8-bit, will you be implementing a dither algorithm for the conversion? This information will help us understand how DiAVC might fit in our playback solutions.

[schweinsz]

Well, probably he would not be willing to release the source code, as he sells SDK licenses, so that would probably be a non-starter.

@schweinsz: Can you elaborate a bit on the future 9/10-bit support? Will DiAVC output be at 8, 10, or 16-bit for these streams? If outputting at 8-bit, will you be implementing a dither algorithm for the conversion? This information will help us understand how DiAVC might fit in our playback solutions.
To the best of my knowledge, the directshow renders supports only 8bits. So the DiAVC will output 8bits to the renderer and the DiAVC SDK could give 16 bits.
I never consider of a dither algorithm now. I am coding the 9/10bits support now.

[madshi]

To the best of my knowledge, the directshow renders supports only 8bits.
madVR supports various 4:2:0, 4:2:2 and 4:4:4 8bit, 10bit and 16bit formats.

[upyzl]

Are you using 64-bit GraphStudio? It should be listed as "DiAVC 64-bit". XhmikosR builds of ffdshow and MPC-HC are available at http://xhmikosr.1f0.de/ but performance is similar to official svn pre-releases.

Many thanks!!:thanks:
As GraphStudio official website only has one exe, I think it's same as CodecTweakTool and Win7DSFilterTweaker (which are one exe handle both 32bit&64bit...)

[upyzl]

well...I'm still in trouble...

after "start"...
http://i.imgur.com/5zM0D.png

How to link it on earth?
The filter can't be inserted, either...

@Stephen R. Savage, I still don't know how you done the benchmark...could you tell me in detail, please?

[Stephen R. Savage]

well...I'm still in trouble...

after "start"...
http://i.imgur.com/5zM0D.png

How to link it on earth?
The filter can't be inserted, either...

@Stephen R. Savage, I still don't know how you done the benchmark...could you tell me in detail, please?

I had some difficulty in getting DiAVC 64-bit to load as well. Due to a coding error, I believe you must first open GraphStudio 64-bit as an administrator, build a graph and insert DiAVC so that it connects, access the options page, close it, and then re-open GraphStudio. Following this, as long as you don't touch the settings page again, DiAVC should load without further issues.

[upyzl]

I had some difficulty in getting DiAVC 64-bit to load as well. Due to a coding error, I believe you must first open GraphStudio 64-bit as an administrator, build a graph and insert DiAVC so that it connects, access the options page, close it, and then re-open GraphStudio. Following this, as long as you don't touch the settings page again, DiAVC should load without further issues.

Er...
The point is, I can't insert DiAVC, though I open GraphStudio 64-bit as an administrator...(build a graph means load a media file, then replace the video decoder by insert another filter, right?)
Others such as FFDshow can be inserted and connected normally...so I can make sure the problem is DiAVC.ax (hope schweinsz can solve it...)

[schweinsz]

The DiAVC x64 free version v1.1 (http://www.di-avc.com/diavc_x64_1.1.zip) is released.
Changes:
dynamic format support
delete some useless code in property page

When you download the diavc_x64_1.1.zip, unpack the .zip to a folder, and run "regsvr32 diavc.ax" in the folder using the cmd.exe, then you can use the diavc.ax x64 version.
Hope the new version is more friendly than the last version.

[schweinsz]

Er...
The point is, I can't insert DiAVC, though I open GraphStudio 64-bit as an administrator...(build a graph means load a media file, then replace the video decoder by insert another filter, right?)
Others such as FFDshow can be inserted and connected normally...so I can make sure the problem is DiAVC.ax (hope schweinsz can solve it...)
Try the new version, could you insert it into the graphstudio?
If not, I can provide you a message-pop version to find the problem tomorrow.

[Stephen R. Savage]

Try the new version, could you insert it into the graphstudio?
If not, I can provide you a message-pop version to find the problem tomorrow.

schweinsz, I just tried DiAVC x64 1.1, and I can not get it to work. When I try to render a media file in GraphStudio, it will not load DiAVC, and I can not access its properties page. I have tried running GraphStudio both as administrator and as a normal user.

http://img851.imageshack.us/img851/8012/diavc64.png (http://imageshack.us/photo/my-images/851/diavc64.png/)

DiAVC does not insert itself when building a graph automatically, and I can not add it manually either.

Edit: After changing the settings of some completely unrelated filter, I can now insert DiAVC x64, but it will still not add itself to an automatically generated graph, and as such I can not gather any performance information, nor can I use it in MPC-HC.

Edit2: I see the following events when trying to insert DiAVC x64 with GraphStudio.

[Fadeout]

Pretty useless right now.

The only way to use it on MPC-HC is starting the player in administrator mode, go in the options, double click DiAVC in the external filters and close the window again. Then you can actually load a video.

Close MPC-HC and you have to repeat the process.

[fairchild]

Pretty useless right now.

The only way to use it on MPC-HC is starting the player in administrator mode, go in the options, double click DiAVC in the external filters and close the window again. Then you can actually load a video.

Close MPC-HC and you have to repeat the process.

That's what I found. It won't load unless I run MPC-HC as admin. (haven't tried other players)

[schweinsz]

Download the DiAVC Msg version (http://di-avc.com/diavc_msg.7z), register it with "regsvr32 diavc_msg.ax", and try to use it. Report the pop message to me. I add some MessageBox in the CDiAVC initialization method.
I tried the diavc x64 v1.1 in graphstudio x64 and MPC-HC x64 all in administrator account and found no problem. I set the DiAVC x64 as the default decoder in the external filter, the MPC-HC will use the DiAVC x64 all along. I try to find the problem with you.

Edit: I uploaded the DiAVC x64 v1.2 (http://di-avc.com/diavc_x64_1.2.zip), some compiler option is changed, hope it can help.

[fairchild]

New working is working now and opening properly with the latest MPC-HC. Thanks!

[Stephen R. Savage]

DiAVC x64 1.2 is now working in GraphStudio. Thanks, schweinsz.

Here's some extended performance information:


CPU: Intel Core 2 Duo T7250 "Merom" (2.00 GHz)
OS: Microsoft Windows Server 2008 R2 Standard

GraphStudio 0.3.2.0 was used for all measurements.
Reported values are the average of three independent trials.
Standard deviation was below 1% of mean for all tested software.
Output accuracy was evaluated visually for each decoder.

1080p, High 4.1, 12.7 Mbps
==========================
x86: 55.0 fps
x64: 54.6 fps

1080p, High 5.0, 9.5 Mbps
=========================
x86: 71.7 fps
x64: 70.1 fps

720p, High 5.0, 11.1 Mbps
=========================
x86: 86.0 fps
x64: 86.3 fps

720p, High 5.0, 2.6 Mbps
========================
x86: 161.9 fps
x64: 161.1 fps

480p, High 3.1, 5.4 Mbps
========================
x86: 300.4 fps
x64: 308.4 fps

480p, High 3.1, 1.8 Mbps
========================
x86: 406.5 fps
x64: 437.0 fps

360p, High 3.0, 1.8 Mbps
========================
x86: 557.8 fps
x64: 594.3 fps


DiAVC x64 doesn't have a measurable performance difference from its 32-bit counterpart, except, rather bizarrely, at extremely low bitrates and resolutions.

[upyzl]

Currently all my problems have been solved.

Thanks!

[Dal]

Hi.
I'm trying to use DiAVC together with DGAVCDecDI, but when i try to open a file, I get this error: Could not create DiAVC Decoder!
I don't really know why, I have run the Registration tool and that produced no error. But when looking at the diavc.ax file, it does not change the modified date when running the registration, only the diavc_mmx does.

Any tips?

Could it be because I have reinstalled windows since I registered DiAVC, and thus need a new machine code and password?

Thanks.

[schweinsz]

Hi.
I'm trying to use DiAVC together with DGAVCDecDI, but when i try to open a file, I get this error: Could not create DiAVC Decoder!
I don't really know why, I have run the Registration tool and that produced no error. But when looking at the diavc.ax file, it does not change the modified date when running the registration, only the diavc_mmx does.

Any tips?

Could it be because I have reinstalled windows since I registered DiAVC, and thus need a new machine code and password?

Thanks.
When you run the DiAVC Registration.exe to activate the DiAVC, assure that the diavc.ax and the diavc_mmx.ax is not used by any applications. Such as, you create a DiAVC filter in the graphedit or the graphstudio, or you try to playback H.264 videoes in player. You don't need a new machine code or password when you re-install windows, but the machine code could change from the original one you sent to me previously. If it is, send the new machine code to me.

[Dal]

When you run the DiAVC Registration.exe to activate the DiAVC, assure that the diavc.ax and the diavc_mmx.ax is not used by any applications. Such as, you create a DiAVC filter in the graphedit or the graphstudio, or you try to playback H.264 videoes in player. You don't need a new machine code or password when you re-install windows, but the machine code could change from the original one you sent to me previously. If it is, send the new machine code to me.
I tried to re-register it, and both ax-files got a new modified date now. But it still does not work. Tried it in DVBViewer and it reports this error: DiAVC init failed.

Sending you my machine code in a PM.

Thanks.

[schweinsz]

I tried to re-register it, and both ax-files got a new modified date now. But it still does not work. Tried it in DVBViewer and it reports this error: DiAVC init failed.

Sending you my machine code in a PM.

Thanks.

Ok, I reply you the new password, your machine code changes from the original one.

[Dal]

Ok, I reply you the new password, your machine code changes from the original one.
It worked like a charm now.
Thanks again.

[IgorC]

There is statistical tool for advanced probabilistic analysis.
http://sjeng.org/ftp/bootstrap.py
http://www.sjeng.org/bootstrap.html
It can be particularly useful for developers and testers to measure small improvements.

[Virtual_ManPL]

It's just me or MPC-HC will just crash (stop responding with Win info about it) on start with DiAVC (all x86-64) when opening some file ?

[schweinsz]

It's just me or MPC-HC will just crash (stop responding with Win info about it) on start with DiAVC (all x86-64) when opening some file ?

What files? Could you share me some among them?

[Virtual_ManPL]

Sure, and sorry for late reply.

Unfortunately for all files...:confused:

http://www.mediafire.com/?isfa4z5q8uuuwgt
http://www.mediafire.com/?l3a1otrai6cgdwf
http://www.megaupload.com/?d=RLJ6QQ1V

EDIT:
It only happens with LAV Splitter so I also post this there, because it's probably splitter issue, not decoder :p

[nevcairiel]

If the decoder crashes, its a decoder issue. It should never crash. :p

[Virtual_ManPL]

I'm not sure what causes it, but works fine with internal MPC-HC splitter :p
Only W7 inform me that MPC-HC stops working and no more info about it

Look here, I reported it on tracker
https://code.google.com/p/lavfilters/issues/detail?id=66

[nevcairiel]

Its the only decoder that crashes, and its closed source, so i cannot check why and how it crashes, so there really isnt anything i can do (Heck, its not even free for me to grab and test)
If schweinsz spots an actual problem with the input data, i can surely try to fix it, but considering that every other decoder that i know of works, i'm still going to bet on the decoder having a small bug there. :)

[schweinsz]

Sure, and sorry for late reply.

Unfortunately for all files...:confused:

http://www.mediafire.com/?isfa4z5q8uuuwgt
http://www.mediafire.com/?l3a1otrai6cgdwf
http://www.megaupload.com/?d=RLJ6QQ1V

EDIT:
It only happens with LAV Splitter so I also post this there, because it's probably splitter issue, not decoder :p

OK, I will download the LAV splitter to test it with the DiAVC.

[Emulgator]

Virtual_ManPL's three files play nicely here under XPSP3 in MPC 6.4.9.1 (no HC) using Haali 1.9.42.1 as splitter and DiAVC 1.2.6 as decoder.

[schweinsz]

I'm not sure what causes it, but works fine with internal MPC-HC splitter :p
Only W7 inform me that MPC-HC stops working and no more info about it

Look here, I reported it on tracker
https://code.google.com/p/lavfilters/issues/detail?id=66
I tested the LAV splitter with the DiAVC on many files in graphstudio x86 and x64, and find it works fine. How to set the LAV splitter as the default splitter for MPC-HC x64?
I import it as external filter and set it as primary filter, but the MPC-HC x64 always selects the Haali splitter.

Besides, I can't download files from both mediafire and megaupload.
Could you upload them to China based netdisk such as http://u.115.com/ or rayfile.com?

[nevcairiel]

You need to add Haali to the external filters as well and set Haali to blocked, its quite .. dominant.

Anyhow, i tested the first sample Virtual_ManPL listed myself, the free DiAVC 64-bit decoder, with LAV Splitter in MPC-HC, and it works just fine.

[Stephen R. Savage]

I also can't reproduce VMPL's issue.

@schweinsz, is there any news about the 9/10-bit support? I know that ETAs are useless esp. regarding software development, but is everything on track, etc.? I know your competitors are "rapidly" working on this.

[schweinsz]

@schweinsz, is there any news about the 9/10-bit support? I know that ETAs are useless esp. regarding software development, but is everything on track, etc.? I know your competitors are "rapidly" working on this.

I have finished more than half of the corresponding codes.

[Virtual_ManPL]

hmm, I use MPC-HC 1.5.3.3678 + LAV Filters 0.32 RC4 + DiAVC 1.2 on Win7 with latest M$ updates & 280.26 nVidia drivers
and MPC always crashing/stop responding on start, after opening some file

did I need maybe some more Visual C++ Redistributables ? because I got only latest Visual C++ 2010 SP1 MFC 64bit

[HJRodrigo]

Have you tried downgrading your Nvidia driver? I installed 280.26 and had the same problem, but with several players. I unintsalled it and installed older drivers and everything started to work again.

[CruNcher]

The new Drivers use a much more aggressive Timing (P-state switching) they got more responsive that way latency was lowered Performance improved it seems that this more stress on the Hardware is causing problems according to user reports many Cards and setups especially under NT 6 are affected, though their are also many Cards and setups that aren't affected. But yeah if this started with the 280.xx Driver you should consider testing with older ones to see if your Setup is affected by these problems.

[Virtual_ManPL]

Thanks for info, see it also on mozillaZine forum :p
Unfortunately, I remember this also happens before I update GPU drivers, but now I will try with 280.28

[HJRodrigo]

From what I read in the forums the previous drivers (275.33) had it too, but slighter than the 280.24...

[schweinsz]

hmm, I use MPC-HC 1.5.3.3678 + LAV Filters 0.32 RC4 + DiAVC 1.2 on Win7 with latest M$ updates & 280.26 nVidia drivers
and MPC always crashing/stop responding on start, after opening some file

did I need maybe some more Visual C++ Redistributables ? because I got only latest Visual C++ 2010 SP1 MFC 64bit
The current DiAVC has no dependency on the vc runtime libarary because I linked it into the .ax when I built it.

[seeker010]

the registration exe takes an hour, and still does nothing. fail.

[LigH]

You seem to be one of very few people with this issue. So it is more probable that your system is damaged or overprotected (e.g. against Registry changes or applications with "unusual behaviour"), or you just miss required prerequisites (like runtime libraries or CPU features), than the registration application being faulty in general (then a lot more people would report such an issue).

Help us helping you; I hope schweinsz will come up with more specific questions to detect the reason of the registration locking up.

[sneaker_ger]

You seem to be one of very few people with this issue. So it is more probable that your system is damaged or overprotected (e.g. against Registry changes or applications with "unusual behaviour"), or you just miss required prerequisites (like runtime libraries or CPU features), than the registration application being faulty in general (then a lot more people would report such an issue).

I also had lots of problems with the activation (Win7 x64). Trying out different versions (Trial and Full) seemed to make everything even worse. At the end I decided to just use a hassle free alternative instead. If you browse through this thread, people with these problems will pop up several times, despite DiAVC being pretty much unheard of outside of doom9. It's a shame, since schweinsz really seems to know his assembly.

[schweinsz]

the registration exe takes an hour, and still does nothing. fail.
Tell me the trouble you met by the e-mail (sales@di-avc.com). I can help you to fix the problem. Does the DiAVC Registration.exe crash?

[schweinsz]

I also had lots of problems with the activation (Win7 x64). Trying out different versions (Trial and Full) seemed to make everything even worse. At the end I decided to just use a hassle free alternative instead. If you browse through this thread, people with these problems will pop up several times, despite DiAVC being pretty much unheard of outside of doom9. It's a shame, since schweinsz really seems to know his assembly.
I feel sorry when I get the news.
I decide to code a simpler registration for the upcomming next release.
If you meet problems on the DiAVC, contact me by the e-mail without hesitate, I always help fix problems.

[madshi]

@schweinsz,

would you mind adding support for DXVA_ExtendedFormat?

http://msdn.microsoft.com/en-us/library/ff563967%28v=vs.85%29.aspx

I've posted detailed information on how to convert the h264 header fields into DXVA_ExtendedFormat in this post:

http://forum.doom9.org/showthread.php?p=1519740#post1519740

madVR makes use of the DXVA_ExtendedFormat information to automatically switch to the correct decoding matrix and primaries. In most cases madVR is able to detect the correct decoding matrix and primaries without DXVA_ExtendedFormat, too, but it'd be nice to have the information available as DXVA_ExtendedFormat, anyway.

[schweinsz]

@schweinsz,

would you mind adding support for DXVA_ExtendedFormat?

http://msdn.microsoft.com/en-us/library/ff563967%28v=vs.85%29.aspx

I've posted detailed information on how to convert the h264 header fields into DXVA_ExtendedFormat in this post:

http://forum.doom9.org/showthread.php?p=1519740#post1519740

madVR makes use of the DXVA_ExtendedFormat information to automatically switch to the correct decoding matrix and primaries. In most cases madVR is able to detect the correct decoding matrix and primaries without DXVA_ExtendedFormat, too, but it'd be nice to have the information available as DXVA_ExtendedFormat, anyway.
I read the struct DXVA_ExtendedFormat and find it is simple. But how can I notify the madVR with the struct?

[nevcairiel]

I read the struct DXVA_ExtendedFormat and find it is simple. But how can I notify the madVR with the struct?

You can put the struct in a VIDEOINFOHEADER2's dwControlFlags field (just cannot use the SampleFormat member, because its space is occupied by (AMCONTROL_USED | AMCONTROL_COLORINFO_PRESENT)

http://msdn.microsoft.com/en-us/library/dd407326%28v=vs.85%29.aspx#extended_color_information

[schweinsz]

You can put the struct in a VIDEOINFOHEADER2's dwControlFlags field (just cannot use the SampleFormat member, because its space is occupied by (AMCONTROL_USED | AMCONTROL_COLORINFO_PRESENT)

http://msdn.microsoft.com/en-us/library/dd407326%28v=vs.85%29.aspx#extended_color_information

Looks it is easy to add the feature.

[madshi]

Agreed, shouldn't be too hard to add, and it shouldn't break anything (if implemented correctly).

[Virtual_ManPL]

dump files
http://www.mirrorcreator.com/files/CUYPX7RC/DUMPs.7z_links

hope it helps diagnosing my crashes

[Toddler Naruto]

How does this Decoder compare to CoreAVC and DivX?

[Stephen R. Savage]

How does this Decoder compare to CoreAVC and DivX?

Performance (8-bit, lossy, CABAC):

1. DiAVC
2. ffdshow
3. CoreAVC / DivX

Hardware acceleration:

DiAVC: No
ffdshow: DXVA
CoreAVC: DXVA, CUDA
DivX: DXVA

Lossless support:

DiAVC: Progressive only
ffdshow: Yes
CoreAVC: Yes
DivX: Don't know

High 10 Profile:

DiAVC: Planned
ffdshow: Yes
CoreAVC: Planned
DivX: No

64-bit binaries:

DiAVC: Yes
ffdshow: Yes
CoreAVC: Yes
DivX: No

Other note-worthy features:

DiAVC: neuron2 support in DGAVCDecDI, dodgy activation mechanism
ffdshow: Free, open-source, video/audio filters, support for other codecs
CoreAVC: You get to pay for each version upgrade... again!
DivX: Has to be installed together with useless DivX software

Basically, DiAVC outperforms other decoders by ~10% in the specific case of CPU decoding on Intel Core-series CPUs for lossy, CABAC-encoded content. Benchmarks for other scenarios can be found in previous pages, usually with inconclusive results. Also, a lot of users report problems with the activation-mechanism in DiAVC.

[Toddler Naruto]

Performance (8-bit, lossy, CABAC):

1. DiAVC
2. ffdshow
3. CoreAVC / DivX

Hardware acceleration:

DiAVC: No
ffdshow: DXVA
CoreAVC: DXVA, CUDA
DivX: No

Lossless support:

DiAVC: Progressive only
ffdshow: Yes
CoreAVC: Yes
DivX: Don't know

High 10 Profile:

DiAVC: Planned
ffdshow: Yes
CoreAVC: Planned
DivX: No

64-bit binaries:

DiAVC: Beta
ffdshow: Yes
CoreAVC: Yes
DivX: No

Other note-worthy features:

DiAVC: neuron2 support in DGAVCDecDI, dodgy activation mechanism
ffdshow: Free, open-source, video/audio filters, support for other codecs
CoreAVC: You get to pay for each version upgrade... again!
DivX: Has to be installed together with useless DivX software

Basically, DiAVC outperforms other decoders by ~10% in the specific case of CPU decoding on Intel Core-series CPUs for lossy, CABAC-encoded content. Benchmarks for other scenarios can be found in previous pages, usually with inconclusive results. Also, a lot of users report problems with the activation-mechanism in DiAVC.

That's great to hear, but will DiAVC ever have Hardware Acceleration?

*is hoping to use his future video card to run most games and videos*

[LigH]

DiAVC has already impressive accelleration by the hardware, regarding the CPU (SIMD). And I believe that's just the intention to avoid too much relying on GPU features. A GPU decoder on a low-featured graphic chipset may even be less performant that a pure CPU decoder on a fast CPU...

[schweinsz]

64-bit binaries:
DiAVC: Beta
ffdshow: Yes
CoreAVC: Yes
DivX: No


I believe the DiAVC 64bits is not beta, it is the free release version.

[Stephen R. Savage]

I believe the DiAVC 64bits is not beta, it is the free release version.

I have corrected my original post. By the way schweinsz, I hadn't checked your website for some time, and I notice there are more files listed on the downloads page. Is there a reason DG Tools must use an older version of DiAVC?

Edit: @Keiyakusha, honestly, I didn't know. Corrected.

[Keiyakusha]

Divx offered DXVA some time ago. did they removed that?

[schweinsz]

Is there a reason DG Tools must use an older version of DiAVC?
Yes, someone reported that there is compatibility problem between the new DiAVC version and the DG tools, but I don't know the reason because I remember that I never changed the SDK.

[Guest]

It's not just the SDK, DGAVCDecDI is sensitive to the decoder's behavior also. DGAVCDecDI has not changed so the problem must have been caused by changes in DiAVC.

I've not yet had time to investigate what has gone wrong.

[Stephen R. Savage]

CoreCodec has just released CoreAVC 3.0 with Hi10p support and planned future 4:4:4 support, along with native high-depth output. They also have added (allegedly) faster ASM. Will post results later when I get my hands on it.

Better get to work, schweinsz! Wouldn't want to fall behind your competition, eh?

[CruNcher]

Yeah more Performant @ least for lossy 8 bit though CoreAVC 3.0 could change that :D

http://img560.imageshack.us/img560/2003/comparex264performancei.png

Divx offered DXVA some time ago. did they removed that?

The DXVA1 implementation of Mainconcept was buggy though the DXVA1 implementation of DivX worked for some time it doesn't work with DXVA2 on SB @ least, might be that it only works with Nvidia or DXVA1 only as Mainconcept also doesn't work with DXVA2 or specifically not with SB.

[nevcairiel]

Yeah more Performant @ least for lossy 8 bit though CoreAVC 3.0 could change that :D

You should really measure performance in FPS, and not in some dodgy cpu load figures.

[Fadeout]

Btw, DiAVC 64bit gave me low CPU usage, but playback wasn't smooth.

[Stephen R. Savage]

CoreCodec has just released CoreAVC 3.0 with Hi10p support and planned future 4:4:4 support, along with native high-depth output. They also have added (allegedly) faster ASM. Will post results later when I get my hands on it.

Better get to work, schweinsz! Wouldn't want to fall behind your competition, eh?

Just got my hands on CoreAVC 3.0. It fails to beat LAV in performance at both 8 and 10-bit. Actually, I think performance has regressed since 2.5.5, but I haven't confirmed that yet. I'm not even sure what these guys do with all the time they spend "developing". :confused:

By the way, I haven't run into choppy playback with DiAVC since one of the early betas added an internal buffer.

Edit: We're still waiting on 10-bit support, schweinsz.

Edit2: Actual benchmark results:


CPU: Intel Core 2 Duo T7250 "Merom" (2.00 GHz)
OS: Microsoft Windows Server 2008 R2 Standard

GraphStudio 0.3.2.0 was used for all measurements.
Reported values are the average of three independent trials.
Standard deviation was below 1% of mean for all tested software.
Output accuracy was evaluated visually for each decoder.

All software used was 32-bit. 10-bit samples were decoded to
8-bit formats in all cases to better represent typical usage
scenarios.

Software Tested:
DiAVC 1.2.6
CoreAVC 3.0
ffdshow (CLSID) r3978
LAV Filters 0.34

1080p, 10-bit, 8750 kbps
========================
CoreAVC: 30.4 fps
ffdshow: 35.7 fps
LAV Filters: 38.2 fps

720p, 10-bit, 1850 kbps
========================
CoreAVC: 60.4 fps
ffdshow: 67.3 fps
LAV Filters: 77.5 fps

720p, 8-bit, 2620 kbps
======================
DiAVC: 157.6 fps
CoreAVC: 133.6 fps
ffdshow: 125.6 fps
LAV Filters: 129.4 fps

480p, 8-bit, 1830 kbps
======================
DiAVC: 390.4 fps
CoreAVC: 340.6 fps
ffdshow: 309.9 fps
LAV Filters: 335.0 fps


It's amazing how far libavcodec has come. It now leads CoreAVC by a staggering 30%, the same margin the initial CoreAVC 1.0 offering lead libavcodec at the time it was released. Nevcairiel is a hero of the workforce.

[CruNcher]

Ehh this is not good :(
DiAVC 64 1.2 worked fine Benchmarking Playback and all but after Benchmarking a lot of crashes all the time now with it i don't hope that your code has the potential to kill the CPU (default intel cooler) ? the only things i did was splitter related installing CoreAVC 2.6.1 though disabling it's splitter and using Lav Splitter like before :( it happened very rapidly the last thing was that a benchmark with DiAVC 64 was very low on the second run it crashed and now it shows artifacting @ playback (better next time monitoring heat) :(
Hmm very strange i try to get rid of everything code related but somehow i have a bad feeling :(

Ah wait the artifacting comes from Lav Audio (actually i change the sync function their when testing ffdshow-quicksync) (uhh man this is always crazy) :P

This really scared me for a second (even knowing that this can come from any code change done in the time running everything):

http://img841.imageshack.us/img841/2264/notgood.png

Playback works but Benchmark crashes i have no idea currently whats going especially as it worked before and i did only very small code changes when it started to crash though 1 was directly related to DiaAVC registering the 32 bit component lets see (unregistered it though 64 bit still crashes when trying to benchmark it).

Hmm Lav Video still works so i guess i can really be sure my hardware is still ok ;)

http://img197.imageshack.us/img197/7296/lavvideostillworks.png

It also almost catches DiAVC 64 1.2 (260 fps) i measured when it was still working :(

Geez this isn't logic (did i triggered somekind of protection mechanism ??)

Playback still works fine but benchmarking fails (though worked before 100%) :(

http://img163.imageshack.us/img163/9209/notlogic.png

Im out of ideas i tried everything registering unregistering killing registry entries killing prefetch getting rid of complete CoreAVC nothing i can't Benchmark DiAVC 64 anymore :P

Or wait maybe mixing Lav Splitter 64 and 32 bit on the same system is the problem ?

PS: Seems i really benchmarked not with Lav Splitter but im sure i did and it didn't crash puhh but benching with MatroskaSplitter.ax works oh man (did you limit the benchmark runs per splitter ;) ? ) (or do i get old but i could swear i used lav splitter in all the previous runs because i had no other *.mkv capable splitter installed until i checked the CoreAVC 2.6 performance and that was after many successful DiAVC 64 1.2 Benchmark runs :( )

Anyways im back on track not with Lav Splitter though but i can survive that for now ;)

[schweinsz]

Ehh this is not good :(
DiAVC 64 1.2 worked fine Benchmarking Playback and all but after Benchmarking a lot of crashes all the time now with it i don't hope that your code has the potential to kill the CPU (default intel cooler) ? the only things i did was splitter related installing CoreAVC 2.6.1 though disabling it's splitter and using Lav Splitter like before :( it happened very rapidly the last thing was that a benchmark with DiAVC 64 was very low on the second run it crashed and now it shows artifacting @ playback (better next time monitoring heat) :(
Hmm very strange i try to get rid of everything code related but somehow i have a bad feeling :(

Ah wait the artifacting comes from Lav Audio (actually i change the sync function their when testing ffdshow-quicksync) (uhh man this is always crazy) :P

This really scared me for a second (even knowing that this can come from any code change done in the time running everything):

http://img841.imageshack.us/img841/2264/notgood.png

Playback works but Benchmark crashes i have no idea currently whats going especially as it worked before and i did only very small code changes when it started to crash though 1 was directly related to DiaAVC registering the 32 bit component lets see (unregistered it though 64 bit still crashes when trying to benchmark it).

Hmm Lav Video still works so i guess i can really be sure my hardware is still ok ;)

http://img197.imageshack.us/img197/7296/lavvideostillworks.png

It also almost catches DiAVC 64 1.2 (260 fps) i measured when it was still working :(

Geez this isn't logic (did i triggered somekind of protection mechanism ??)

Playback still works fine but benchmarking fails (though worked before 100%) :(

http://img163.imageshack.us/img163/9209/notlogic.png

Im out of ideas i tried everything registering unregistering killing registry entries killing prefetch getting rid of complete CoreAVC nothing i can't Benchmark DiAVC 64 anymore :P

Or wait maybe mixing Lav Splitter 64 and 32 bit on the same system is the problem ?

PS: Seems i really benchmarked not with Lav Splitter but im sure i did and it didn't crash puhh but benching with MatroskaSplitter.ax works oh man (did you limit the benchmark runs per splitter ;) ? ) (or do i get old but i could swear i used lav splitter in all the previous runs because i had no other *.mkv capable splitter installed until i checked the CoreAVC 2.6 performance and that was after many successful DiAVC 64 1.2 Benchmark runs :( )

Anyways im back on track not with Lav Splitter though but i can survive that for now ;)

The DiAVC x64 is without any protection-related mechanism.
I benchmarked the DiAVC x64 with the LAV splitter just then and find the same problem: the graphstudio x64 crashed. I will try to find the problem.

[CruNcher]

The DiAVC x64 is without any protection-related mechanism.
I benchmarked the DiAVC x64 with the LAV splitter just then and find the same problem: the graphstudio x64 crashed. I will try to find the problem.

Yeah i use MPC-HC standalone 64 filters currently because of the DiAVC 1.2 Lav Splitter issues


Here is a sample which shows a hit on both Lav Video and DiAVC 1.2, CoreAVC 3.0 seems to get better around with it (and or those are splitter decoder interaction problems) :)

http://img11.imageshack.us/img11/8761/evanl.png

On this Youtube 4K release also CoreAVC 3.0 does better

http://img196.imageshack.us/img196/70/lifeinthegarden.png


Yeah i use MPC-HC standalone 64 filters currently because of the DiAVC 1.2 Lav Splitter issues

[Stephen R. Savage]

By the way schweinsz, are there any plans to finish the lossless MBAFF support? I recall that you almost had it finished, but it crashed when used in DirectShow or something. I hope you get around to this some day, as it is a notable and kind of silly omission.

[CruNcher]

@schweinsz
i found several apple encoded bitstreams that seem to have performance issues it seems (it could be also that the splitter is actually not working correctly with the decoder on those but that would be strange)

There is a really interesting tendency on those streams CoreAVC 3.0 Performance is going down on them and Lav Video can win on those though DiAVC still stays @ it's first position, only 1 Apple encoded stream the one above (evan.mov) shows that CoreAVC 3.0 can go ahead of DiAVC on those :)

Except those 2 small Performance issues with the 4k and Apple sample i couldn't yet find any where DiAVC 8 bit loses vs Lav Video or CoreAVC 3.0, but im still on the search ;) :)

Found another one also again Apple Encoded Bitstream


PS: That CoreAVC 3.0 is sometimes faster could have todo with it's Memory Performance but im not 100% sure

http://img263.imageshack.us/img263/401/lettersfrom.png

[schweinsz]

By the way schweinsz, are there any plans to finish the lossless MBAFF support? I recall that you almost had it finished, but it crashed when used in DirectShow or something. I hope you get around to this some day, as it is a notable and kind of silly omission.

Do you mean the PAFF lossless? I finished it in the last year, and the code is right. but the compiler gives wrong decoded results. I don't know the reason.

[Stephen R. Savage]

Do you mean the PAFF lossless? I finished it in the last year, and the code is right. but the compiler gives wrong decoded results. I don't know the reason.

Yes, this is what I meant. I was hoping you had solved the miscompilation issue.

Btw, CRUNCHER, could you stop posting those huge images please? It causes page stretching and makes this thread hard to read in general.

@CruNCher, it causes stretching on my copy of Firefox 6 in 960x1200 (half fullscreen). Generally, though, it can cause posts to get lost esp. if they're between two of your image-loaded posts :(

[CruNcher]

Hmm on which Browser and Resolution it causes stretching problems ?

PS: Found a crash condition

http://img269.imageshack.us/img269/9831/diavccrash.th.png (http://img269.imageshack.us/img269/9831/diavccrash.png)

Here is the corrupted bitstream that causes the crash http://www.mediafire.com/download.php?la0dfgrr4c8zkwu

[Toddler Naruto]

DiAVC has already impressive acceleration by the hardware, regarding the CPU (SIMD).

And I believe that's just the intention to avoid too much relying on GPU features.

A GPU decoder on a low-featured graphic chipset may even be less performance that a pure CPU decoder on a fast CPU...

Oh, I see, I had no idea it was better that way.

[Gser]

I was wondering what kind of IDCTs DiAVC and CoreAVC use?

[LigH]

Because MPEG4-AVC hardly uses the DCT = Discrete Cosine Transformation at all (maybe except for the High Profile?), but an Integer Transformation instead ... usually none.

[Stephen R. Savage]

schweinsz, this sample with MKV ordered chapters + linked segments doesn't play correctly with DiAVC. Seeking creates a distorted image, with an apparent error in the pitch.

Files:

fileExternal.mkv:
MultiUpload: http://www.multiupload.com/A3VZML0CBU
MegaUpload: http://www.megaupload.com/?d=QDG2KOXW
DepositFiles: http://depositfiles.com/en/files/tzjybigow

fileMain.mkv:
MultiUpload: http://www.multiupload.com/SSY5KX5Q6L
MegaUpload: http://www.megaupload.com/?d=13K2KLCX
DepositFiles: http://depositfiles.com/en/files/bumiheetk


DiAVC presents:
http://img828.imageshack.us/img828/2122/diavcborked.th.png (http://imageshack.us/photo/my-images/828/diavcborked.png/)

Correctly decoded picture:
http://img846.imageshack.us/img846/692/correcte.th.png (http://imageshack.us/photo/my-images/846/correcte.png/)

Steps to reproduce:
1. Install Haali Media Splitter. Ensure that "try to open linked files" is enabled.
2. Disable any intermediate processing filters like ffdshow, VSFilter, etc.
3. Open file "fileMain.mkv". Video from "fileExternal.mkv" should begin playing.
4. Hit "Page Down" to advance by one chapter. Image should now be distorted.

Inserting an intermediate filter like ffdshow raw video filter or VSFilter results in the correct image being displayed. LAV Video Decoder and ffdshow Video Decoder both produce the correct image.

[Pat357]

I bought DiAVC yesterday (paid using Paypal), but haven't got any e-mail back containing my registration code.

Is this an automated procedure are went something wrong ?
Who can I contact about my registration code ?

[LigH]

The developer of DiAVC, schweinsz, is present here. He will read your reply probably soon. I believe he has to confirm your payment manually, so don't get panic if it delays for a few days.

[schweinsz]

I bought DiAVC yesterday (paid using Paypal), but haven't got any e-mail back containing my registration code.

Is this an automated procedure are went something wrong ?
Who can I contact about my registration code ?

I replied you allready. I check my e-mail every day except I can't connect to the internet. I seldom can't connect to the internet. You can send e-mail to me at sales@di-avc.com.

[Pat357]

I replied you allready. I check my e-mail every day except I can't connect to the internet. I seldom can't connect to the internet. You can send e-mail to me at sales@di-avc.com.

OK, I just send out email with PayPal receipt as proof of my payment to sales@di-avc.com.
Thanks in advance !
-
pat357

[schweinsz]

OK, I just send out email with PayPal receipt as proof of my payment to sales@di-avc.com.
Thanks in advance !
-
pat357

I received your payment for some days. I replied you twice and it seems that you never get it. You should check your spam e-mail.

The current registration system don’t support the user-defined user name, you must provide me the register name by running the DiAVC Registration.exe.

Setup the DiAVC and run the DiAVC Registration.exe, copy the machine code and send it to me.
See the webpage and find where you can get the registrater name.
http://di-avc.com/buynow.html

[CruNcher]

@schwinsz
any news about bugfixes with lav splitter and at least that 1 bitstream crash i found more strange beahaviour which seems parsing related but i wanted to wait for those fixes to see if those might fix the other issues as well before reporting :)

[schweinsz]

@schwinsz
any news about bugfixes with lav splitter and at least that 1 bitstream crash i found more strange beahaviour which seems parsing related but i wanted to wait for those fixes to see if those might fix the other issues as well before reporting :)

I am busy on algorithms for HEVC. so it is delayed.

[LigH]

I am busy on algorithms for HEVC.

:scared: Eh? HEVC (http://en.wikipedia.org/wiki/HEVC) — the to-be-expected H.265? Interesting. Are there any encoders available already? ... We'll be curious. :sly:

[schweinsz]

:scared: Eh? HEVC (http://en.wikipedia.org/wiki/HEVC) — the to-be-expected H.265? Interesting. Are there any encoders available already? ... We'll be curious. :sly:

Only the reference encoder/decoder (http://hevc.kw.bbc.co.uk/trac/browser/trunk)is available now to the best knowledge of mine.
The HEVC is its normal name, I don't know its future name, but it perhaps isn't named as H.265, because I find that the MPEG strongly takes against that it uses a ITU type name.

[CruNcher]

Ligh dark tested this Samsung BBC proposal you remember ;) ?

[hajj_3]

Interesting to see work being done on HEVC. I wonder whether hardware decoding chips on current cpu's and gpu's could cope with hardware decoding, i highly doubt it. Judging my DS's review of it no cpu will be able to decode it which means we won't be able to use HEVC for quite some time.

Are you able to even do any work on HEVC yet, isn't it better for you to wait until february when the draft standard is supposed to be ready?

[schweinsz]

Interesting to see work being done on HEVC. I wonder whether hardware decoding chips on current cpu's and gpu's could cope with hardware decoding, i highly doubt it. Judging my DS's review of it no cpu will be able to decode it which means we won't be able to use HEVC for quite some time.

Are you able to even do any work on HEVC yet, isn't it better for you to wait until february when the draft standard is supposed to be ready?

I currently don't code the encoder/decoder of HEVC, I just try to incorporate my algorithms into the standard.

[CruNcher]

good luck schweinsz looking @ what you did with DiAVC im pretty sure they cant ignore your ideas for improvement :)

[benus]

I plan to add 9, 10bits support for the DiAVC next.

I have finished more than half of the corresponding codes.

First quote is from 22nd July, second from 18th of august. Come on man, it's been three months.
I want to see a comparison between and DiAVC and CoreAVC for decoding 10-bit material.

Could you be decent and say how much time it'll take to accomplish a project.

Regards and thanks for your hard work.

[Asmodian]

This is not a good way to ask this question.

No matter how politely you ask if you don't provide a pay check or have a working relationship you do not get to be annoyed by delays or demand status updates.

schweinsz is a decent guy even if he never offers us a timeline or gets back to DiAVC. :mad:

[hajj_3]

he won't be making much from DiAVC but if his algorithms are used in h.265 then he'll be paid a small amount for every chip and license made, he could make millions! Clearly it is in his interest to create those algorithms than to focus on DiAVC. The draft for h.265 is in a few months so i'm sure he'll be concentrating on this then.

[mariush]

He can dream... if it's like h264 where there's hundreds of patents, his patents/algorithms will be a drop in a lake of patents so probably the majority of the money will go to the big companies.

Still, better than nothing I suppose.

[schweinsz]

Recently I received two payment from USA and Japan. But they both never provide the right machine code to me. I reply them by e-mail to request the machine code but never get reply. If you see my post, please contact with me by sales@di-avc.com.

[LigH]

;) Those are just incognito donations. ;)

[TheRyuu]

DiAVC Registration.exe just sits in some sort of inf. loop and nothing ever pops up.

As much as I enjoy being punished by invasive drm on software which I've bought, the fun sort of wears out after a while.

[schweinsz]

DiAVC Registration.exe just sits in some sort of inf. loop and nothing ever pops up.

As much as I enjoy being punished by invasive drm on software which I've bought, the fun sort of wears out after a while.

The DiAVC Registration.exe crashes. Give me your e-mail, I will help you fix the registration problem.

[madshi]

Bug report:

It seems that DiAVC does not set the "IMediaSample2::GetProperties" flag "AM_VIDEO_FLAG_REPEAT_FIELD". This is an important piece of information which is needed to reliably detect a 3:2 pulldown pattern for IVTC algorithms. Without this flag, soft-telecined content will look like 2:2 while hard-telecined content will look like 3:2.

[TheRyuu]

The DiAVC Registration.exe crashes. Give me your e-mail, I will help you fix the registration problem.

Sent one to sales@di-avc.com.

[XMEN3]

There is not support in megui for this tool.
If i try to load an index to crate script it gives me a popup error like this:
http://img249.imageshack.us/img249/3120/errorjr.png (http://img249.imageshack.us/i/errorjr.png/)

So to load the index i must rename DGavcdecDi.dll to DGavdecNV.dll and put it on dgindexnv folder.
Also diavc.ax is in the same folder.
Doing so i can create a script but when i try to load it megui crashes with this error:
Firma problema:
Nome evento problema: BEX
Nome applicazione: megui.exe
Versione applicazione: 1.0.2068.0
Timestamp applicazione: 4ee68152
Nome modulo con errori: diavc.ax_unloaded
Versione modulo con errori: 0.0.0.0
Timestamp modulo con errori: 4ceec48f
Offset eccezione: 7384c768
Codice eccezione: c0000005
Dati eccezione: 00000008
Versione SO: 6.1.7601.2.1.0.1296.17
ID impostazioni locali: 1040
Informazioni aggiuntive 1: fa66
Ulteriori informazioni 2: fa6696398de2b9f98383d7a3bf5c3ea1
Ulteriori informazioni 3: d1ed
Ulteriori informazioni 4: d1edbcc00c19a40f9b56bb439c959571


The script is loaded fine on virtualdub.
LoadPlugin("C:\Program Files (x86)\MeGUI-x86\tools\dgindexnv\DGDecodeNV.dll")
DGSource("E:\test\test.dgi",fieldop=0)
LanczosResize(624,352) # Lanczos (Sharp)

[schweinsz]

There is not support in megui for this tool.
If i try to load an index to crate script it gives me a popup error like this:
http://img249.imageshack.us/img249/3120/errorjr.png (http://img249.imageshack.us/i/errorjr.png/)

So to load the index i must rename DGavcdecDi.dll to DGavdecNV.dll and put it on dgindexnv folder.
Also diavc.ax is in the same folder.
Doing so i can create a script but when i try to load it megui crashes with this error:
Firma problema:
Nome evento problema: BEX
Nome applicazione: megui.exe
Versione applicazione: 1.0.2068.0
Timestamp applicazione: 4ee68152
Nome modulo con errori: diavc.ax_unloaded
Versione modulo con errori: 0.0.0.0
Timestamp modulo con errori: 4ceec48f
Offset eccezione: 7384c768
Codice eccezione: c0000005
Dati eccezione: 00000008
Versione SO: 6.1.7601.2.1.0.1296.17
ID impostazioni locali: 1040
Informazioni aggiuntive 1: fa66
Ulteriori informazioni 2: fa6696398de2b9f98383d7a3bf5c3ea1
Ulteriori informazioni 3: d1ed
Ulteriori informazioni 4: d1edbcc00c19a40f9b56bb439c959571


The script is loaded fine on virtualdub.
LoadPlugin("C:\Program Files (x86)\MeGUI-x86\tools\dgindexnv\DGDecodeNV.dll")
DGSource("E:\test\test.dgi",fieldop=0)
LanczosResize(624,352) # Lanczos (Sharp)


What can I do to support the meGui?

[XMEN3]

What can I do to support the meGui?
About the problem with megui, neuron2 said that will not develope anymore DGavcdecDI from alpha stage and also said that will refound the donation...
So, as i've paid your decoder to use with it i'm asking the same thing to you,if is possible...
Thanks

[benus]

I would like to ask schweinsz a question.

Since you have mentioned planning a release of 10-bit Diavc, would you mind informing us whether you are going to proceed with it.
If it's yes what would be approximate time of launch.

Thanks for your work.

[schweinsz]

I would like to ask schweinsz a question.

Since you have mentioned planning a release of 10-bit Diavc, would you mind informing us whether you are going to proceed with it.
If it's yes what would be approximate time of launch.

Thanks for your work.

I pause the work in the later of the last year and concentrate on many kinds of algorithms of the next generation video coding standard (HEVC related). I believe I will be back and code to finish the 10-bits support of the DiAVC in this May.

[hajj_3]

have any of your algorithms been accepted in the HEVC spec?

[benus]

I pause the work in the later of the last year and concentrate on many kinds of algorithms of the next generation video coding standard (HEVC related). I believe I will be back and code to finish the 10-bits support of the DiAVC in this May.

Thank you for your answer. It is a blessing to be informed at any given time. Good look with both of the projects.

[schweinsz]

have any of your algorithms been accepted in the HEVC spec?
No, it is very hard to adopt an algorithm that is from a person who is not belong to any big company.
I am inventing algorithms to make a private standard that is competitive to the HEVC or better than it.

[hajj_3]

you are making a whole new codec as a competitor to HEVC? Like VP8 vs H.264? If so, surely that is too much work for 1 man? You might be better off just selling your ideas to google to implement in VP9 or something as i find it hard to see any company wanting to use your codec over that of google/mpeg-la's as any patent disputes can be funded by these rich companies.

[schweinsz]

you are making a whole new codec as a competitor to HEVC? Like VP8 vs H.264? If so, surely that is too much work for 1 man?
Most of algorithms are ready now. I have the new transform kernel, the new luma/chroma interpolation filter, new arithmetic coder, new deblocking archetecture algorithm, new context models for MV and coefficients coding, new inter pretiction method and new block splitting symetics. I am only lack of the new intra coding method.

You might be better off just selling your ideas to google to implement in VP9 or something as i find it hard to see any company wanting to use your codec over that of google/mpeg-la's as any patent disputes can be funded by these rich companies.
Is there the channel to sell my algorithms to google?

[Emulgator]

That sounds awesome.
At my 2006 editing system (no DGDecNV possible) diavc still makes my day. And now this.
Forgive my curiosity, schweinsz, if I may ask out of respect: At which university did you study ?

[hajj_3]

Is there the channel to sell my algorithms to google?

DARK SHIKARI is the main developer for x264, google contributes so i'm sure he could ask google for you.

[CruNcher]

you are making a whole new codec as a competitor to HEVC? Like VP8 vs H.264? If so, surely that is too much work for 1 man? You might be better off just selling your ideas to google to implement in VP9 or something as i find it hard to see any company wanting to use your codec over that of google/mpeg-la's as any patent disputes can be funded by these rich companies.

he wrote a complete optimized H.264 Decoder in a relative short time compared to his competitors (commercial/open source) (though still not as many features) ;)
Though it's sad that people like Schweinsz and Neiromaster work in the same direction that MPEG moves in and don't try other grounds like Snow and Dirac and going into efficient parallel algorithms on GPUs :)

[IgorC]

Is there the channel to sell my algorithms to google?
Google is looking for new generation codecs (patent free of course).

Xiph foundation is preparing to develop a new Daala codec.
http://wiki.xiph.org/Daala
http://git.xiph.org/

You can join IRC channel irc.freenode.net #VP8 and #daala

[Midzuki]

...
Xiph foundation is preparing to develop a new Daala codec.
http://wiki.xiph.org/Daala
http://git.xiph.org/

Hmmm, that's interesting, :thanks: for the new info.

As a "casual end-user", I only hope the project won't be a container-locked ffmpeg2daala :rolleyes: ,
nor receive a rather-unimaginative name like x-daala :p

[schweinsz]

Google is looking for new generation codecs (patent free of course).

Xiph foundation is preparing to develop a new Daala codec.
http://wiki.xiph.org/Daala
http://git.xiph.org/

You can join IRC channel irc.freenode.net #VP8 and #daala

I read their web-page just then. As they want to get a patent-free codec, I believe they can't use the reference frame more than 2 and the 10bits technology. The multiple reference frame is patented by Thomas Wiegand for a company many years ago while the 10bits technology is strictly patented in almost all big countries by Toshiba. I ever read the two patents files downloaded from the eu patent web-site.

[CruNcher]

Don't forget Intel acquired Real Networks NGV just recently with all patents ;)

[BetaBoy]

Don't forget Intel acquired Real Networks NGV just recently with all patents ;)
It's gonna be fun to see how Intel unravels this.

[BetaBoy]

Google is looking for new generation codecs (patent free of course).

Xiph foundation is preparing to develop a new Daala codec.
http://wiki.xiph.org/Daala
http://git.xiph.org/

You can join IRC channel irc.freenode.net #VP8 and #daala
While I applaud the effort, its an uphill battle crowded and complex as the video patent pools are. It's gonna be a miracle if WebM will even hold up no less any new potential patent free video codec.

[CruNcher]

yep if people don't move away from MPEG no chance, though those who tried in the past weren't really successful either and completely vanished or adapted ;)
The People working on Daala tough know exactly what MPEG is and how to use their principles and modify them so that they wont be the same on paper even 10 bit as 10 bit is a principle that's hardly patentable so Toshiba could say nothing against a different way of implementation :)
Google on the other side has so much power they could really pull it of buying ON2 was the first step but building up on the same principals is wrong doing research is the right way pushing VP9 in a complete different direction if Google can't do it i really wonder how Xiph could even if i have the biggest respect for those working behind Daala, though maybe not having this power @ all in the end is going to succeed ;)
Though the last i know of who tried to knock out of MPEG where DivX and where are they today, their last rescue in the end was to adapt and being a good implementer i mean i highly credit it to Gej even thinking of creating something unique (their is this rumor of DivX own Codec Research in the early days) after steal eh borrow MPEGs work but in the end he landed exactly their just implementing (though just implementing is maybe to low credited he made it into Hollywood's circles being the provider for their next Digital Push, though not with their Video Research but DRM Research) ;)

Though a Video Codec isn't the same complexity as a Browser it's far far higher even a very good high paid team just implementing is happy with small gains in month it's not like creating a browser @ version 1 and being @ version 18 in some years ;)
Thus i also have big respects for Guys like Jarret ,Loren, Neiromaster Picard and Schweinsz :)
And moving away from implementing to create something new is always a great evolution for such guys and i applaud them for trying it :)
Neiromaster and Schweinsz would be for sure a great addition to Daala :)

The Most exciting was to see what the BBC have came up with with Dirac and of course Niedermayer with Snow in the last 10 years man that was impressive :)

[hajj_3]

schweinsz, this looks like it might be an early version of VP9, thought you might be interested: https://gerrit.chromium.org/gerrit/#change,17840

[Stephen R. Savage]

schweinsz's latest excursion sounds like another pipe dream like that "temporal super resolution" business way back when this was in the alpha stages. I only hope he comes back down to earth soon so we can get updates for 10-bit support.

On another note, it seems neuron2 has canceled DGAVCDecDI.

[Pat357]

schweinsz's latest excursion sounds like another pipe dream like that "temporal super resolution" business way back when this was in the alpha stages. I only hope he comes back down to earth soon so we can get updates for 10-bit support.

On another note, it seems neuron2 has canceled DGAVCDecDI.
Could be for a lot of reasons, but I think that one of them might be the incompatibility from his app. with the latest DiAVC versions.
Only the older DiAVC v 1.2 can be used with his DGAVCDecDI, all newer versions are no longer compatible.

That means unless Scheinz fixes this issue, we will never be able to use DGAVCDecDI with Hi10P stuff.
Also support for High 4:4:4 (inclusive lossless) with Avisynth 2.6 (Y24) would never be supported.

[Emulgator]

I would be thankful for a hint,
can't seem to find a working registry entry for DiAVC's frameserver settings.
(WinXP32 SP3)

Right now I got some poor and underexposed 1280x720x25p on 50p AVCHD cam footage to edit.

If I preview in MPC using DiAVC and DiAVC "Properties" Window:

A: Crushed shadows if I leave levels set to "Auto".

B: Still underexposed, but shadows are not crushed too much anymore
when I manually set Levels as follows: Input PC Range, Output TV Range.

Now I'd like to do frameserving using DGAVCDecDI and I would like to use the same preset, but how ?

The frameserved result unfortunately gets crushed blacks.
The histogram borders come out close to a rectangle with a strong peak at Y=0.
The DGAVCDecDI chain seems to assume "Input/Output Level: Auto/Auto"
so obviously does not read from the GUI settings.

A decoding workaround:
Sony Vegas Pro 10 still decodes this stream with a 0-255 histogram with a slight peak at Y=0 or 1,
and not too much crushed shadows, so I decode in Vegas for now.

[schweinsz]

I would be thankful for a hint,
can't seem to find a working registry entry for DiAVC's frameserver settings.
(WinXP32 SP3)

Right now I got some poor and underexposed 1280x720x25p on 50p AVCHD cam footage to edit.

If I preview in MPC using DiAVC and DiAVC "Properties" Window:

A: Crushed shadows if I leave levels set to "Auto".

B: Still underexposed, but shadows are not crushed too much anymore
when I manually set Levels as follows: Input PC Range, Output TV Range.

Now I'd like to do frameserving using DGAVCDecDI and I would like to use the same preset, but how ?

The frameserved result unfortunately gets crushed blacks.
The histogram borders come out close to a rectangle with a strong peak at Y=0.
The DGAVCDecDI chain seems to assume "Input/Output Level: Auto/Auto"
so obviously does not read from the GUI settings.

A decoding workaround:
Sony Vegas Pro 10 still decodes this stream with a 0-255 histogram with a slight peak at Y=0 or 1,
and not too much crushed shadows, so I decode in Vegas for now.

The DiAVC treats the input/output range in the directshow filter, and the DiAVC SDK only output the information about it.
The DGAVCDecDI tools use the DiAVC SDK directly, so setting the property page is useless.

[benus]

I pause the work in the later of the last year and concentrate on many kinds of algorithms of the next generation video coding standard (HEVC related). I believe I will be back and code to finish the 10-bits support of the DiAVC in this May.

Since we have June already I would like to ask whether you are any closer to launch a 10 bit DIAVC decoder.
I would not like you to think I am putting any sort of pressure on you.
I am just looking for the information.
:o

[benus]

I pause the work in the later of the last year and concentrate on many kinds of algorithms of the next generation video coding standard (HEVC related). I believe I will be back and code to finish the 10-bits support of the DiAVC in this May.

Hi schweinsz.

It is July now and still we have not heard from you about the progress of your works over new DiAVC 10-bit version.
Could you be so kind and let us know if you plan to implement the code or perhaps you dropped the idea.

Regards

[schweinsz]

Hi schweinsz.

It is July now and still we have not heard from you about the progress of your works over new DiAVC 10-bit version.
Could you be so kind and let us know if you plan to implement the code or perhaps you dropped the idea.

Regards
Sorry that I cancel the plan to finish the DiAVC 10-bit version. I am working on a new private video coding codec that will be similar performance with the HEVC. I am working on intra-coding currently.

[hajj_3]

could you give us some more indepth information on your own codec. Did you contact google, mozilla or xiph about selling your codec patents to them?

[schweinsz]

could you give us some more indepth information on your own codec. Did you contact google, mozilla or xiph about selling your codec patents to them?

Currently I don't contact with anyone in google, mozilla or xiph and I don't know anyone or any contact information in the three company/organization. Anyway, I am very glad to sell my codec and all the patents to them.
I ever read the xiph's daala and google's on-going vp9. Because I have an entirely new architecture/algorithms, I find it is hard to integrate most of my algorithms into their architecture, so I decide to work out a totally new codec with my algorithms.

[dukey]

schweinsz,
I don't think the buy section on your website is working correctly. Paypal says it can't load correctly because the window is in a frame. I guess you could get around this by, not using frames on your website, or opening the buy window inside a new tab/window.

[an3k]

Seriously:

First there is a guy developing a decoder and most people are only blaming him for his "slow decoder", "crappy code" and whatever else. Then they notice they were kinda wrong but never excused themselves.

Then neuron2 completely removed DGAVCIndex and published DGAVCIndexDI instead, forcing users to additionally buy a DiAVC Decoder license. Then suddenly both projects are dead but instead of telling interested customers (since both of them sell software licenses) that the project is dead both websites are still the same, offering a license.

I know that developing nearly everything yourself can be very boring and exhausting over a long time but if you sell your software you have to support it for at least a year after you sold it to a customer. What would you say when you just bought a license and notice some month later by accident that the software you bought a license for is dead?

Why did i "upgrade" from DGAVCIndex to DGAVCIndexDI? Because DGAVCIndex processes ONLY the first file, even if you opened multiple ones and because it's extremely slow. Multi-Threading? 64-Bit?

[Guest]

Then neuron2 completely removed DGAVCIndex I withdrew it because libavcodec became unusable for accurate random access due to regressions and my trouble report was ignored. Additionally the "hall of shame" incident disinclined me to want to work further with people that treated me that way.

and published DGAVCIndexDI instead, forcing users to additionally buy a DiAVC Decoder license. I didn't force anything. I offered an additional option for decoding. At the time it looked like DiAVC was here to stay. To be honest, I don't know the state of DiAVC at this time but as long as DGAVCDecDI still works with the shipped DiAVC binary there is no reason to withdraw it.

Then suddenly both projects are dead DGAVCDecDI is not dead. It still functions properly and ships with a compatible diavc.dx. If you are saying that DiAVC licenses are no longer available, then this is the first I have heard of it and if it is indeed true, then I will withdraw DGAVCDecDI.

but instead of telling interested customers (since both of them sell software licenses) that the project is dead both websites are still the same, offering a license. So you say that DiAVC licenses are in fact still available?! Then what are exactly are you complaining about?

Why did i "upgrade" from DGAVCIndex to DGAVCIndexDI? Because DGAVCIndex processes ONLY the first file, even if you opened multiple ones and because it's extremely slow. Multi-Threading? 64-Bit? If you are not happy with DGAVCDecDI then request a refund and I'll be happy to oblige. My flagship product is DGDecNV. I made DGAVCDecDI to offer some kind of option to people without an nVidia card. At some point DiAVC broke backwards compatibility with DGAVCDecDI and was apparently not willing to work with me to correct that and stop that from happening again. Tracking a moving target is difficult and time consuming, and not economically practical for a product with a very small user base and my very low license fee. So I decided to leave DGAVCDecDI as it is and ship an older known compatible diavc.dx binary. The shipped version still works and some people without other options continue to find it useful.

[an3k]

I withdrew it because libavcodec became unusable for accurate random access due to regressions and my trouble report was ignored. Additionally the "hall of shame" incident disinclined me to want to work further with people that treated me that way.Didn't knew about that. What's the "hall of shame" incident? just found a japanese website saying something about GPL.

I didn't force anything. I offered an additional option for decoding. At the time it looked like DiAVC was here to stay. To be honest, I don't know the state of DiAVC at this time but as long as DGAVCDecDI still works with the shipped DiAVC binary there is no reason to withdraw it.Well, since DGAVCIndex got removed it is kind of forcing customers to buy another license :) Currently licenses for DiAVC can still be bought. Last update is from May 31st, 2011, see DiAVC Changelog (http://www.di-avc.com/changeLog.html)
I'm currently waiting for a reply from the dev about machine id change but if nothing happens should i inform you?

DGAVCDecDI is not dead. It still functions properly and ships with a compatible diavc.dx. If you are saying that DiAVC licenses are no longer available, then this is the first I have heard of it and if it is indeed true, then I will withdraw DGAVCDecDI.See neuron2.net Forum (http://neuron2.net/board/viewtopic.php?f=6&t=118#p1717). Since you deleted my account there (or wiped the whole forum) i have to ask here: Where did you said that it's alpha?
I count a project as dead if there is (likely) no further development; that doesn't mean that the latest release doesn't work :)

So you say that DiAVC licenses are in fact still available?! Then what are exactly are you complaining about?That i had to buy a license for a decoder i don't want but have to use because DGAVCIndex got removed completely and DGAVCIndexDI only works with DiAVC.
I really don't know if the last working libavcodec offers more or at least the same decoding features as DiAVC 1.2.2 does but keeping the support for it would have been great.

If you are not happy with DGAVCDecDI then request a refund and I'll be happy to oblige. My flagship product is DGDecNV. I made DGAVCDecDI to offer some kind of option to people without an nVidia card. At some point DiAVC broke backwards compatibility with DGAVCDecDI and was apparently not willing to work with me to correct that and stop that from happening again. Tracking a moving target is difficult and time consuming, and not economically practical for a product with a very small user base and my very low license fee. So I decided to leave DGAVCDecDI as it is and ship an older known compatible diavc.dx binary. The shipped version still works and some people without other options continue to find it useful.On my computer i use DGIndexNV (i was an early adopter - before your license generator was online ;) and i really like that tool but since i got a spare Dell Server with two Xeon QuadCores i want to use that one for decoding too. Sadly it has only PCI-X and PCI-E x8 slots so afaik i cannot add a nVIDIA GPU and that forces me to use DGAVCIndex(DI).

[Guest]

I'm currently waiting for a reply from the dev about machine id change but if nothing happens should i inform you? Yes, please do.

Where did you said that it's alpha? In the Readme that ships with DGAVCDecDI.

I count a project as dead if there is (likely) no further development; that doesn't mean that the latest release doesn't work I never promised to continue developing anything in perpetuity.

[an3k]

In the Readme that ships with DGAVCDecDI.Oh, yes, right. I must have overread that one word :)

I never promised to continue developing anything in perpetuity.And that's not what i request but buying a license and noticing by accident some month later that development of the software has stopped is just annoying.

But perhaps there will be an (at least minor) updated version after you merged DGIndex, DGAVCIndex and DGIndexNV into one program - as you indicated in the neuron2.net forum ;)

[Guest]

Be reasonable! You think merging support for CUVID, libavcodec, and DiAVC would be a minor update?

Anyway, now I think you are being disingenuous, as you stated you initially made a donation to get DGDecNV long ago, so your claim of making a donation for DGAVCDecDI only to find after a month later that I don't intend to develop it further is a fabrication. And for what purpose?

As I have told you several times, if you are unhappy then send me your info and I will issue you a refund. Then I will not have to deal further with your silliness.

[an3k]

Be reasonable! You think merging support for CUVID, libavcodec, and DiAVC would be a minor update?

But perhaps there will be an (at least minor) updated version of the DGAVCIndex part after you merged DGIndex, DGAVCIndex and DGIndexNV into one program - as you indicated in the neuron2.net forum ;)

This is what i meant, thought it was clear enough before!

Anyway, now I think you are being disingenuous, as you stated you initially made a donation to get DGDecNV long ago, so your claim of making a donation for DGAVCDecDI only to find after a month later that I don't intend to develop it further is a fabrication. And for what purpose?I made ONE donation for DGIndexNV long time ago. Period! I never said i made another donation for DGAVCIndexDI. I said i bought a license for DiAVC! You know that DGAVCIndexDI without DiAVC is completely useless?! Please read more carefully!

As I have told you several times, if you are unhappy then send me your info and I will issue you a refund. Then I will not have to deal further with your silliness.Wow! Are you serious? Insulting a customer who tried to help you where possible and violating forum rules (Rule #4 to be exact)?! Anyway, this is the wrong place for discussing the problems between us and since you're already a moderator i sadly have to contact doom9.

[Guest]

I cannot be held responsible for whether or not DiAVC continues development or not. This is what I mean by your silliness. Why don't you go rag on them? You've had the benefit of my DGDecNV product now for years for a measly $15 and yet here you are ragging on me. I explained the situation to you as clearly and openly as I could but you won't let it go. And why the profanity that I had to remove? You set the tone buddy.

[an3k]

I cannot be held responsible for whether or not DiAVC continues development or not. This is what I mean by your silliness.Example:nVIDIA now wants a license fee for software that implements OpenCL/CUDA directly. You buy a license. Two month later you hear by accident that nVIDIA is going a complete new way and that OpenCL/CUDA is no longer maintained/supported. But they still ask for a license fee and didn't posted a notice about the discontinuation on the website.It is kind of the same just with the exception that you develop DGIndexNV yourself and didn't had to buy a license for it.

I absolutely know that you cannot be held for DiAVC and i know that there is no guarantee that the kind of software like DiAVC, DGIndex, AviSynth, etc. is maintained till the end of the world but you also stopped developing DGAVCIndexDI. First completely removing DGAVCIndex and then stopping development of the replacement. All these three things are connected.

Why don't you go rag on them?Isn't this the main thread of DiAVC? If schweinsz is dead, on vacation or just to busy with developing something else that he doesn't read this and replies - what else should i do? If he doesn't send me the updated license code because his software thinks the machine code changed for no reason (many users have that issue) i'll get him to refund. Perhaps his decoder is nice, perhaps not. I don't know.

You've had the benefit of my DGDecNV product now for years for a measly $15 and yet here you are ragging on me.True and the price is ok for that. I paid 50 € for AnyDVDHD and i get lifetime updates/upgrades. Higher price but they also release much more updates. Same for many other software out there! But it doesn't mean that you can instantly stop development/support for a software just because the user only paid once. To be fair you at least should inform users on your website so they can decide themselves if they want to buy a license for a more or less soon dying software. It's about fairness!

I explained the situation to you as clearly and openly as I could but you won't let it go.Then you absolutely misunderstood me. I was just discussing with you about that stuff. I never was trying to get my money back from you. I never tried you to begin development of DGAVCIndex again - it would be nice, yes but i know that its the developers decision and that is very hard to change ;)

And why the profanity that I had to remove? You set the tone buddy.That was just to express how i see the whole situation. As i wrote there were some users kind of "bullying" schweinsz for his initial postings. Then the force for users to buy a DiAVC license or keep the not well working old DGAVCIndex version. Then the issue with stopped development but no information on the website - neither on yours nor on DiAVCs. In short: "WHAT THE FREAK" was not directed to anyone here! It was not my intention to insult someone and I'm sorry if i did.

...
...
...

And thank you for the refund but you really don't have to do that. I'm using DGIndexNV and i like to keep using it (i could even without rejecting ...) but I'm fair and thus I'm rejecting your refund! Perhaps we just got one's wires crossed and i hope that this post explains it a bit better and that we aren't each others deadly enemy :)

[LigH]

It is indeed a pity that schweinsz decided to be ... "less active" with DiAVC. For any reason he may have. I believe one of them was professionally developing H.265 software. Yes, five.

But if you still prefer (or are stuck to) software-only decoding, there are also still alternative ways. FFMS2 is still being developed, trying to get rid of "insanity" in the multi-threaded decoding. And DirectShow is the last hope anchor, which can be successful if used wisely, carefully tweaking filters and merits to ensure that the graph does nothing else but decoding with optimal quality, especially without unexpected additional filtering.

[LoRd_MuldeR]

an3k, the fact that one tool (DGAVCIndex) happened to be released as a Freeware neither gives you any guarantees for future updates nor does it give you any guarantees that such an update would still be available as Freeware. And of course having been a user of one tool (DGAVCIndex) doesn't give you the "right" to use other Payware tools, such as DGDecNV or DGAVCIndexDI, for free. Nobody forced you to buy any of those! You could have continued using DGAVCIndex as-is or you could have used an alternative. Instead you decided to buy DGAVCIndexDI+DiAVC. So what exactly are you complaining about? For what I understand, the existing DGAVCIndexDI still does work with the "compatible" version of DiAVC. Right? So, unless there is a particular problem, what is your concern? You bought a software and it works. I don't think anybody gave you a guarantee for "free lifelong updates" when you bought your DGAVCIndexDI+DiAVC licenses. And that's not unusual at all! For example: If you buy a Microsoft Office today, you get the "current" version and that's it. You might get some "security" updates/fixes now and then for free, but for next "major update" you have to pay again! Or stick with your "old" version...

[Guest]

Actually, I do offer free lifetime updates, if I create any! I don't guarantee to make any update or future versions of my donation-ware, but if I do they will always be free to existing licensees.

But your point stands, one cannot expect that development must necessarily continue just because a working version is still sold.

[schweinsz]

If you are not happy with DGAVCDecDI then request a refund and I'll be happy to oblige. My flagship product is DGDecNV. I made DGAVCDecDI to offer some kind of option to people without an nVidia card. At some point DiAVC broke backwards compatibility with DGAVCDecDI and was apparently not willing to work with me to correct that and stop that from happening again. Tracking a moving target is difficult and time consuming, and not economically practical for a product with a very small user base and my very low license fee. So I decided to leave DGAVCDecDI as it is and ship an older known compatible diavc.dx binary. The shipped version still works and some people without other options continue to find it useful.
I never change the interface of the DiAVC for a very long time. Actually I don't know why it becomes not compatible with the DGAVCDecDI. Now I concentrate on a new video coding project that is with better coding efficiency than the HEVC. but I still maintain the DiAVC project.

[Guest]

Well, schweinsz, with all due respect, nothing changed in DGAVCDecDI and then one day things stopped working with one of your new releases. We've discussed this but you were not interested in resolving it, just kept saying nothing changed. This is why I had to stop my development.

Don't you recall writing this to me:

"I notice the same problem; perhaps it is because that I change some compiler options of the DiAVC."

What ticked me off is that you just started releasing versions that you knew broke backward compatibility without discussing it with me or informing me. That told me you didn't care about DGAVCDecDI. So why should I care about DiAVC? I thought we had a good partnership going and that we could do great things together. Then you started not caring, and I suppose it is because you moved on to your new encoder.

but I still maintain the DiAVC project

Then you can release a latest version that is compatible with DGAVCDecDI just as version 1.2.2 was? If not, then what does "maintain" actually mean to you?

[schweinsz]

Well, schweinsz, with all due respect, nothing changed in DGAVCDecDI and then one day things stopped working with one of your new releases. We've discussed this but you were not interested in resolving it, just kept saying nothing changed. This is why I had to stop my development.

Don't you recall writing this to me:

"I notice the same problem; perhaps it is because that I change some compiler options of the DiAVC."

What ticked me off is that you just started releasing versions that you knew broke backward compatibility without discussing it with me or informing me. That told me you didn't care about DGAVCDecDI. So why should I care about DiAVC? I thought we had a good partnership going and that we could do great things together. Then you started not caring, and I suppose it is because you moved on to your new encoder.
I don't know what I changed in detail. I only know that the interface functions are not changed. I only add some more features into the DiAVC.
"I notice the same problem; perhaps it is because that I change some compiler options of the DiAVC."
You should notice the "perhaps".



Then you can release a latest version that is compatible with DGAVCDecDI just as version 1.2.2 was? If not, then what does "maintain" actually mean to you?
For example, I fix a problem in the GetMachineCode.exe for some computers that it is hard to get their harddisk serial number in recent days.

[Guest]

I don't know what I changed in detail. I only know that the interface functions are not changed. If you have two versions of DiAVC and one has the problem and one doesn't, it is a simple matter to determine the source code changes and isolate which one caused the problem. But without your source code, there's no way anyone other than you can do that. You're a really smart guy, why isn't this obvious to you? Maybe because you don't care? I certainly hope it is not because you have no source code control!

I only add some more features into the DiAVC.
"I notice the same problem; perhaps it is because that I change some compiler options of the DiAVC."
You should notice the "perhaps". The point is you knew about the problem but weren't willing to do anything about it. You just silently broke other apps depending on continued backward compatibility.

For example, I fix a problem in the GetMachineCode.exe for some computers that it is hard to get their harddisk serial number in recent days. OK, great. But you have no time to figure out how you broke backward compatibility? As it stands, I can't make your SDK and current decoder version work together, whereas things were fine with a previous decoder version. This should be a nightmare for someone distributing a decoding library, but you appear totally unconcerned.

[schweinsz]

If you have two versions of DiAVC and one has the problem and one doesn't, it is a simple matter to determine the source code changes and isolate which one caused the problem. But without your source code, there's no way anyone other than you can do that. You're a really smart guy, why isn't this obvious to you? Maybe because you don't care? I certainly hope it is not because you have no source code control!

I DO have no strict source code control for the DiAVC source code. Before each release, I test all hundreds bitstreams for correctness and I assume everything is OK. As I don't change the interface, I assume that DG tools work fine.
I often compress the source code to a .7z file and store it to another place, but not in every release. I have limited experiences on the project management. This is a lesson for me.

[LoRd_MuldeR]

I DO have no strict source code control for the DiAVC source code. Before each release, I test all hundreds bitstreams for correctness and I assume everything is OK. As I don't change the interface, I assume that DG tools work fine.
I often compress the source code to a .7z file and store it to another place, but not in every release. I have limited experiences on the project management. This is a lesson for me.

Sounds like you should definitely do yourself (and your business partners) a favor and have a look at Git!
http://git-scm.com/book

Developing a non-trivial software without a proper version control system can get really messy, even if you work alone. But as soon as a team/collaboration is involved, it becomes pretty much impossible to handle.

Regression tests are extremely helpful too. But what if a test fails (one that used to work) and you need to figure out what change exactly broke it? Without fine-grained version control you're doomed...

[Guest]

I DO have no strict source code control for the DiAVC source code. Ouch. That's pretty shocking.
Before each release, I test all hundreds bitstreams for correctness and I assume everything is OK. As I don't change the interface, I assume that DG tools work fine. But do you actually test the interface by running an application? It should be very simple to make sure your SDK interface is still working. Just because the interface doesn't change that doesn't mean it still functions correctly given other changes to the decoder. It also confuses me a little because you say you were aware of the problem.
I often compress the source code to a .7z file and store it to another place, but not in every release. Even copying every release would have served you well here. Why wouldn't you copy every release? Disk size is not a concern these days.

[dukey]

I think it's a real shame because this codec is very fast, I think has a _lot_ of potential. But it feels unfinished. The directshow part probably works more by accident than design I think. Seeing the source for the directshow part that schweinsz sent me it's no surprise. It's really a crazy mess full of Messagebox debugging and other insanities.

[schweinsz]

Ouch. That's pretty shocking.
But do you actually test the interface by running an application? It should be very simple to make sure your SDK interface is still working. Just because the interface doesn't change that doesn't mean it still functions correctly given other changes to the decoder. It also confuses me a little because you say you were aware of the problem.
Even copying every release would have served you well here. Why wouldn't you copy every release? Disk size is not a concern these days.
I have a diavc.exe that is the build version of the DiAVC SDK.
After each building, I will run the diavc.exe to verify if the interface works fine. It is always fine.
I have many .7z backups that is only labeled by date, for example, 20100719_1.7z.

[schweinsz]

I think it's a real shame because this codec is very fast, I think has a _lot_ of potential. But it feels unfinished. The directshow part probably works more by accident than design I think. Seeing the source for the directshow part that schweinsz sent me it's no surprise. It's really a crazy mess full of Messagebox debugging and other insanities.
But the directshow filter works fine. There is only one issue that I can't fix. As I code the directshow filter, I have very limited knowledge on this, so it is started from scratch.
Anyway, I am not with good code style.
As seldom peoples use the DiAVC, I stopped its development and concentrate on the next generation video coding algorithm development. I try to finish a private codec that is comparable with or better than the HEVC.

[dukey]

Every time I paused video playback became unplayable. I would love to use this codec, but it was kind of a deal breaker. You might find that a lot more people would use diavc if it was a bit more stable, and you had a nicer website. As I pointed out, people can't even buy it unless they go directly to the link. Paypal refuses to work in frames.

You should open source the directshow part of your filter.

[schweinsz]

Every time I paused video playback became unplayable. I would love to use this codec, but it was kind of a deal breaker. You might find that a lot more people would use diavc if it was a bit more stable, and you had a nicer website. As I pointed out, people can't even buy it unless they go directly to the link. Paypal refuses to work in frames.

You should open source the directshow part of your filter.

OK, you can get the source code of the DiAVC transform filter source code here (http://di-avc.com/diavcfilters.7z).

[schweinsz]

Every time I paused video playback became unplayable. I would love to use this codec, but it was kind of a deal breaker. You might find that a lot more people would use diavc if it was a bit more stable, and you had a nicer website. As I pointed out, people can't even buy it unless they go directly to the link. Paypal refuses to work in frames.

You should open source the directshow part of your filter.

Does paypal work in a div? I plan to change the frames to div.

[Disabled]

Probably not. Why not do it like everyone else, send them directly to paypal? I'd assume paypal also has a guide on how to implement their service into your website.

[ChibiBoi]

Did schweinsz go on vacation? I ordered DiAVC three days ago and I haven't gotten a password yet :-/ I anxious to test this out! :)

[LoRd_MuldeR]

Did schweinsz go on vacation? I ordered DiAVC three days ago and I haven't gotten a password yet :-/ I anxious to test this out! :)

Well, he said that he now concentrates "on a new video coding project".

[ChibiBoi]

Well, he said that he now concentrates "on a new video coding project".

Yeah, I read that :-/ well I wanted to use DiAVC 1.2.2 with DG tools, but then I read the forums that he and neuron2 were having a disagreement with some things. But schweinsz hasn't been on doom9 since 2 weeks ago, so I'm hoping he's just gone on vacation and he'll get back to me soon LOL

[schweinsz]

Did schweinsz go on vacation? I ordered DiAVC three days ago and I haven't gotten a password yet :-/ I anxious to test this out! :)
OK, you will get it tonight.

[schweinsz]

Well, he said that he now concentrates "on a new video coding project".
Yes, the intra-coding related program is finished. I will open it recently. All critical algorithms are invented basically before this march. I just implement it one part by one part.

[hajj_3]

Yes, the intra-coding related program is finished. I will open it recently. All critical algorithms are invented basically before this march. I just implement it one part by one part.

We already have VP9 now though, i don't see much point in your codec being created as google's codec is going to be great and will have hardware decoding built into chips.

[schweinsz]

We already have VP9 now though, i don't see much point in your codec being created as google's codec is going to be great and will have hardware decoding built into chips.
Do you mean that a new codec is useless? I can do better than VP9, perhaps better than HEVC.

[LigH]

:eek: I'm looking forward to your proof. You certainly chose a respectable challenge. :cool:

[war59312]

schweinsz, I for one, hope you kick Google's ass! Good lucky buddy!

Though it would be wise to get a better website already. ;) Frames suck!!

And you need to update your COMPARISON page quickly. Should test with a modern i7 CPU and against CoreAVC 3.0 and LAV Filter.

[[SLiM]D12]

Looking to purchase this program, but i couldn't find a support contact, so i hope somebody here can assist. From what i gather on the resgistration page, i'm to use my "Machine Code" as my registration name, but for some reason when launching the DiAVC registration window, my machine code is blank.

I've searched thru the 72 pages here unsuccessfully and i'm hoping somebody can help me out.

Thanks

Edit: Seems that it's a Windows 8 issue, tried compatibility settings, but couldn't get a machine code to appear -- is there really no way for this to work in windows 8?

[kypec]

[SLiM]D12 AFAIK DiAVC had reached its end of life and support many months ago. Currently, there are much better alternatives available, which are free and very actively developed. Please have a look at LAV Filters thread (http://forum.doom9.org/showthread.php?t=156191) and I'm sure you'll be satisfied.

[schweinsz]

D12;1666617']Looking to purchase this program, but i couldn't find a support contact, so i hope somebody here can assist. From what i gather on the resgistration page, i'm to use my "Machine Code" as my registration name, but for some reason when launching the DiAVC registration window, my machine code is blank.

I've searched thru the 72 pages here unsuccessfully and i'm hoping somebody can help me out.

Thanks

Edit: Seems that it's a Windows 8 issue, tried compatibility settings, but couldn't get a machine code to appear -- is there really no way for this to work in windows 8?
Download the GetMachineCodeWoHDD (http://di-avc.com/getMachineCodeWoHDD.ZIP) and run it to get the machine code and send it to me, then when you get the password, run the GetMachineCodeWoHDD to input the password to activate it.

[samf88]

I bought a license for DiAVC a few days ago as I was hoping to make use of DGAVCDecDI... expecting my license key to be delivered automatically. Still haven't heard anything and have filed a claim with PayPal... not impressed!

[Asmodian]

DGAVCDecDI has been abandoned and I believe the newest versions of DiAVC are not compatible with it.

[LigH]

I'd recommend to try L-SMASH Works if you don't have supported hardware for DG's decoders (NV / IM); in this case, regarding DGAVCDecDI, maybe ask for a refund.

[videoh]

I bought a license for DiAVC a few days ago as I was hoping to make use of DGAVCDecDI... expecting my license key to be delivered automatically. Still haven't heard anything and have filed a claim with PayPal... not impressed! I assume you mean you didn't get the license from DiAVC. Graft responds very quickly to such problems for DG tools.

[samf88]

Yeah, it's DiAVC I haven't heard from. He really shouldn't be selling his software if he is unable to offer the product in a decent timeframe...

Never heard of L-SMASH. Since I got rid of my NVIDIA graphics card I've missed DGDecNV so was hoping for a suitable alternative as DGAVCDec doesn't work very well with my files and FFMS2 seek issues are a big problem.

[LigH]

L-SMASH Source (http://forum.doom9.org/showthread.php?t=167435): A kind of "successor" and fork from FFmpegSource2, but also including the L-SMASH splitter for ISO Media containers (MP4, QT MOV, etc.) in addition to libav.

Latest builds by the_weirdo (https://www.dropbox.com/sh/3i81ttxf028m1eh/AAABkQn4Y5w1k-toVhYLasmwa?dl=0) since VFR maniac doesn't provide binaries regularly anymore.

This is a software-only decoder, but very fast (especially when multi-threaded) and reliable, has less issues with non-MKV containers (which was the biggest limitation of FFMS2).

And there is even one more alternative: DSS2Mod (http://ge.tt/2QzHlt02/v/0), a modified "DirectShowSource2" (based on the one shipped with Haali Media Splitter) which can use LAV Filters natively instead of installed DirectShow filters. The Google Code project (http://code.google.com/p/xvid4psp/downloads/detail?name=DSS2%20mod%20%2B%20LAVFilters.7z&can=2&q=) documents this different download link for the latest version.

[samf88]

Thanks LigH, will be sure to check those out! :)

[travolter]

Im using old netbooks and Im trying to use the most efficient video decoder for H264.

I was reading all the thread and the superiority of DiAVC over coreAVC... but that was in 2011.

What are the most efficient H264 SOFTWARE decoders in 2024 for old machines that does not support CUDA/DXVA/hardware accel.?

[Emulgator]

DiAVC.

[travolter]

Im looking for DiAVC x64 free version. My netbooks need magical powers ;)
https://thumbs4.imagebam.com/e7/e6/cc/MEVODS8_t.png (https://www.imagebam.com/view/MEVODS8)
The DiAVC page is gone now. Download is not working on the wayback machine
http://web.archive.org/web/20141128205926/http://di-avc.com/

Anyone have link to any of these free versions of diavc_x64? Maybe someone conserve it on an old HD :)
diavc_x64.zip
diavc_x64_1.1.zip
diavc_x64_1.2.zip

thanks ;)

[travolter]

Just a question for everyone that used DiAVC in the past.

I can only connect DiAVC with mpc-hc videoplayer.... and its simply amazing!!! but I wonder why other players like potplayer or mpc-be fails to connect (I have registered/selected the filter to decode h264/AVC, and set merit to prefer... but currently the only player that works is mpc-hc.

Why? Maybe DiAVC coder tested only with that player?

There is any trick or something that Im missing?

[Emulgator]

That setting may be the systemwide merit setting.
The other players may have their own preference chains which may overrule system wide merits,
but player-dependent you can maybe edit these chains again to honor system-wide merits.
WinDSFilterTweaker 6.5 is helpful to see what is going on.
http://www.codecguide.com/windows7_preferred_filter_tweaker.htm