DG NV tools

[Karyudo]

I've just had a frustrating--but ultimately rewarding--experience troubleshooting an "error" while using DGIndexNV that I thought I'd document for posterity, just in case somebody else needs it.

I recently built a new Windows 7 (64-bit) machine, and installed DGIndexNV, build 2009, under 'Program Files (x86)'. I then went online and generated a license for my new machine, and added it to my 'license.txt' file using copy and paste.

After this was all done and I'd installed AviSynth 2.58 and VirtualDub 1.9.9, I tried to run an AviSynth script I'd created and run before on my old machine, and got this error:

VirtualDub Error [ <-- title of window ]

Avisynth open failure:
DGSource: Invalid license.
(\\HOMESERVER\Users\Karyudo\_Video Render Projects\final.avs, line 20)

A reboot didn't help (not too surprisingly, but one never knows with Windows...). Neither did any of the other editing of the .avs that I tried. So I contacted neuron2, who helped me troubleshoot the issue:

• DGIndexNV appeared to work fine when creating new projects, which implied that the 'license.txt' file was correct and working as it should.

• Irritatingly, everything also worked fine if I started an entirely new DGIndexNV project from the new computer, and then opened the resulting .dgi with DGSource().

• The same .avs that worked, edited to point to a .dgi created on my old machine, did NOT work. This indicated that something was wrong with my .dgi files.

After I sent off some sample .dgi files, neuron2 very quickly found my issue: The path to the license file stored in the (human-readable) .dgi was pointing to 'Program Files' rather than to 'Program Files (x86)' as it needed to be under my 64-bit OS!

Once I added those few characters, all was well. All done, and in less than 8 (overnight) hours from initial contact to successful resolution.

Moral(s) of the story? Neuron2 provides excellent support of his products... and I'm clearly sort of an idiot.

[Guest]

and I'm clearly sort of an idiot. Don't be too hard on yourself, because the info on that string is buried in an obscure appendix of the users manual.

Thanks for the report that may benefit others.

[Karyudo]

I'm being hard on myself because I'm usually the sort of guy who thinks, "Moron! Don't you know the info on that string is buried in an obscure appendix of the users manual??"

[Clumpco]

Sorry for the lousy "report", I know how unhelpful that is, but I can't actually reproduce the error at will. 'Rather hoping to "ring some bells" maybe..
But wait; you mean with DGMultiSource() I can start two instances?! I'm missing something then, cause I use DGMultiSource() exclusively, and I always get this on second instances: Failed to create video decoder (100).
I'm admin on an XP SP3 with the latest patches, Comodo FW, Antivir Premium using a GeForce 8600 GT, Driver version 6.14.11.9621.

I think that I might have found the circumstances to re-create this error...
Build 2009 - if you encode from an DGIndexNV avs using MeGui and then leave MeGui open, when you restart DGIndexNV and open another file you get "Failed to create video decoder (100)".
Close MeGui and try again and all is fine.

Does this ring any bells for the original poster?

Incidentally I sometimes get the same error at the start of the second pass - this is relatively new but I wouldn't like to say if it is since a DGI update or a MeGui one.

[Guest]

Sounds like something for Zathor to look into as nothing has changed in DGDecNV that would account for that.

[Clumpco]

Sounds like something for Zathor to look into as nothing has changed in DGDecNV that would account for that.
Yes, smells of an unreleased file handle to me (or whatever is the equivalent for a frameserver).

[WasF]

Does this ring any bells for the original poster?
No, I don't use MeGUI at all (I make my own .bat with x264 options, edit cuts with VDub, and build AVS Trim() instructions - and Adobe Audition's .scp script - from VDub's .vcf file using a small AWK script of my invention). I like getting my hands dirty :rolleyes:

neuron2's explanation actually made perfect sense (lack of memory in my video card, I have 256MB).
The only issue was "corruption" of current instance when a second one was launched. But I haven't faced the issue ever since, so I'm rather thinking my system was somehow instable at the time, and this has nothing to do with DG Tools.

[tormento]

Is there a way to disable audio demux from a m2ts thru the GUI, thus creating the dgi only?

[Guest]

I could say RTFM...

Audio/Audio Demux/Clear All/Done

[tormento]

It was just a bit hidden ;)

I was asking that as I usually transcode dts and ac3 to lower bitrates to create mkv from BD. Could be nice to interface DGNV to eac3to and let it make all the work...

[Clumpco]

Audio/Audio Demux/Clear All/Done

Feature request: Checkboxes ! Much faster than clearing/selecting when there are multiple streams.

[tormento]

If we are in the request day, I'd like a mpls parser too...

[Guest]

MPLS parsing is already on the to-do list. GUI stuff I don't get too excited about.

I'm working on next generation security to fight the serial gens and exe crackers. I want to at least make it a bit hard for them, so they get bored putting so much work into a $15 program and give up. :)

[Mixer73]

3. DGDecNV had PureVideo deinterlacing and GPU resizing. Unless you really need 3, there is no compelling reason to use DGDecNV *for MPEG2*.

I'm considering DGDecNV myself, honestly I should just donate for all the good works you've done, I must have done thousands of videos with your tools built into various GUIs over time.

My question is this: will the GPU decode faster than software for HD MPEG2 streams? I have an i7 965, maybe I'm just impatient but I want it to be faster in the indexing stage ;)

[Guest]

My question is this: will the GPU decode faster than software for HD MPEG2 streams? I have an i7 965, maybe I'm just impatient but I want it to be faster in the indexing stage ;) Indexing does not use the GPU, so it can provide no boost. Decoding alone will likely not be faster on the GPU.

Possible gain comes from transcoding (decoding plus encoding), not simply decoding, as described several times. Offloading decoding to the GPU can free up CPU for encoding.

But any performance gain is secondary to the other advantages, also described several times: reliable random access, GPU deinterlacing, GPU resizing/cropping, correct decoding of AVC and VC1, consistent integrated solution to AVC/MPEG/VC1.

[tormento]

But any performance gain is secondary to the other advantages, also described several times: reliable random access, GPU deinterlacing, GPU resizing/cropping, correct decoding of AVC and VC1, consistent integrated solution to AVC/MPEG/VC1.
One of the feature I used most is GPU deinterlacing.

One of the feature I need most is GPU denoising which is a killer for CPU but should be easily parallelized with CUDA or so. Film grain is something abused that increase bitrate for free.

Neuron: do you have the skills to implement some advanced noise filters such as MCTD or MDegrain into DGNV?

[Renzz]

I've hit a problem using dgindexnv 2009 trying to index a particular MPEG2 TS file. It's reporting "malformed GOP" (twice, then it stops), however, if I run the same file through dgindex, it indexs it fine. For what it's worth, Videoredo also reports no errors in the stream, either when editing it or using QuickStreamFix.

I've cut the offending part out and it can be found at http://www.mediafire.com/?y5j2nziqmgj . It would be good to know if this is a bug in dgindexnv, or a real corruption in the stream that the other tools miss.

Thanks in advance.

[Guest]

I've hit a problem using dgindexnv 2009 trying to index a particular MPEG2 TS file. It's reporting "malformed GOP" It's caused by GOPs like this:

SH
GOP
I
B
B
P
B
B
P
B
B
SH
P
B
B

It's unusual to have a sequence header (SH) appear before a P frame like that within a GOP. I detect boundaries for sorting into display order using the SHs in DGIndexNV, while in DGIndex I use the I frames (can't remember why I changed that). I actually have an old scribbled note here warning about this but I didn't worry because I never saw this in the wild. I'll have to change it to use the I frames as in DGIndex. It may take a few days.

Thanks for your report.

[Renzz]

Thanks for the quick response, it's much appreciated.

[Renzz]

Yes, smells of an unreleased file handle to me (or whatever is the equivalent for a frameserver).

I too get the "Failed to create video decoder" error intermittently using MeGui. It's much worse using dgmultisource, but less frequent when using Cuvidserver. I originally put this down to my ancient Nvidia card (8600 with 256MB) but I recently installed a 9800GT with 1GB of memory - but I still get the problem. It's also worse when your source is H264, and almost non-existant when it's MPEG2.

MeGui is definitely holding onto something - if you run GPU-Z (http://www.techpowerup.com/downloads/SysInfo/GPU-Z/) and monitor memory usage of the GPU, you can see that whenever you open the AVS in MeGui, memory goes up, but is never fully released. However, I only ever get to around 400MB before hitting the problem, so it must be some other resource issue. Odd that it's worse with DGMultiSource as opposed to Cuvidserver - I would have expected them to be the same, maybe?

[CruNcher]

@Neuron2
Does nvcuvid.dll (VP2) support Mpeg-2 4:2:2 Studio Broadcast streams ?

[Guest]

Currently no, but funny you should ask. I am working with Nvidia on that right now.

[aegisofrime]

neuron2, I'm thinking of donating for your DG NV Tools and I have some questions I hope you can help me with.

The Purevideo deinterlacing function looks great, and that's what I guess I will be using most. The thing is my work computer is also my gaming computer. I'm currently using a Radeon 4870 and need an nVidia card. I have come up with 2 choices

1) Get a Fermi (GTX 470 is my first choice)

2) Continue using my Radeon 4870 and get a cheap GT210/GT220 and use it alongside with the 4870 purely for Purevideo.

My question is, is the purevideo deinterlacing quality dependent on the card or are there differences? Also, do you know if your software will work in scenario 2? I'm aware that only Windows 7 supports heterogeneous video card support. I'm currently on vista and will have to upgrade to 7 for that, but I hope to find out whether it will work first.

Thanks!

[squid_80]

I've been running ATI and nVidia cards in the same machine since windows XP, no need for windows 7.

[LigH]

Additional SH in GOP ... wild guess: To support decoders catching a new video stream after DVB channel switching?

[Guest]

My question is, is the purevideo deinterlacing quality dependent on the card or are there differences? I don't know the answer to that. You'd have to ask Nvidia. I do know that it will be fine with the 210/220.

Also, do you know if your software will work in scenario 2? I'm aware that only Windows 7 supports heterogeneous video card support. I'm currently on vista and will have to upgrade to 7 for that, but I hope to find out whether it will work first. Yes, that scenario is supported.

[Guest]

Additional SH in GOP ... wild guess: To support decoders catching a new video stream after DVB channel switching? No, because an SH doesn't make a P frame into an I frame, so that can't improve seeking at that point. Anyway the GOP is very small to begin with.

[Guest]

* The previously hidden transport stream trim functionality is now exposed in
the GUI. A progress dialog for the trim operation has been added. Performance of the
file write operation has been improved. The functionality is now more robust for
writing proper packet boundaries. 192 byte transport packet streams are now
properly supported. Refer to the users manual for important usage notes and
caveats.

* Added a CLI avs template command: -at file.

* Fixed the Disable Display function for double rate deinterlacing. The operation
of this function has been made more robust.

* Fixed a problem causing some AAC streams to not be detected.

* Fixed a problem causing some teletext streams to be detected as audio.

* Added an error popup for non-4:2:0 streams (until supported by Nvidia).

* Fixed a problem in indexing of MPEG2 streams that caused some unusual but legal
GOPs to be considered malformed and cause an abort of the indexing.

http://neuron2.net/dgdecnv/dgdecnv.html

[Guest]

BTW, my 64-bit system is up and running now. Clean install and activate on bare metal with only a Win7 upgrade disk is quite a challenge, but I was up to it. :)

Development of 64-bit DGDecNV is begun.

[Clumpco]

BTW, my 64-bit system is up and running now... Development of 64-bit DGDecNV is begun.

Yay! ... and :sad: I can't afford an I7 and Win7 64 bits for the moment :)

[aegisofrime]

I don't know the answer to that. You'd have to ask Nvidia. I do know that it will be fine with the 210/220.

Yes, that scenario is supported.

Thanks for the reply :) If I do get a Fermi I will report back on its quality.

I've been running ATI and nVidia cards in the same machine since windows XP, no need for windows 7.

Thanks. I'm aware that heterogeneous is supported on XP as well, but I greatly prefer 7 to XP.

[Clumpco]

Version 2010 now successfully avoids treating Teletext as audio :-)
Unfortunately it still seems to detect BBC HD ac3 2/0 as 3/2. But no biggie.

[Guest]

Unfortunately it still seems to detect BBC HD ac3 2/0 as 3/2. Did you give me a stream for that? If so, what was the name? If not, please do so.

[djloewen]

I've got a video that isn't behaving optimally. It's an extra from a blu-ray disc, encoded at 720x480 (and it should be 4:3 but it's apparently not flagged as such, it appears to even display on a television in a slightly stretched AR). eac3to detected it as "480p24/1.001 (3:2) with pulldown flags". When it's indexed using DGindexNV 2010 it reports "Frame Rate: 23.976024 fps" but it seems to be counting 30fps worth of frames. As a result, when I encode the video I end up with ~22mins of audio and ~28mins of video (30fps played back at 24fps). Small sample here:
http://www.megaupload.com/?d=PW53CQ2O

Any help would be appreciated.

[Mixer73]

Development of 64-bit DGDecNV is begun.

64bit Vista here for testing of any builds you wish, although I am only using DGDecNV in a very limited way.

[Guest]

eac3to detected it as "480p24/1.001 (3:2) with pulldown flags". I don't see any soft pulldown in the stream you linked.

When it's indexed using DGindexNV 2010 it reports "Frame Rate: 23.976024 fps" but it seems to be counting 30fps worth of frames. The stream is indeed coded as 23.976 fps. Regarding your second point, I cannot tell anything without the muxed stream to compare the audio. It sounds unlikely.

Please post a link to a larger M2TS source sample direct from the bluray that I can use to duplicate the problem.

[Clumpco]

Did you give me a stream for that? If so, what was the name? If not, please do so.Not recently, no. However I suspect that the .ts that it did it on was corrupt, after "fixing" DGNV detected it correctly as 2/0.

I shell get back to you after more investigation.

[djloewen]

I don't see any soft pulldown in the stream you linked.

The stream is indeed coded as 23.976 fps. Regarding your second point, I cannot tell anything without the muxed stream to compare the audio. It sounds unlikely.

Please post a link to a larger M2TS source sample direct from the bluray that I can use to duplicate the problem.

Here's a 100mb sample of the original m2ts file:
http://www.megaupload.com/?d=MKOFDA0W

Here's the full m2ts file, in case it helps (763mb):
[link removed]

This particular episode (there are 4) is 20:17, but the video ends up being 25:21.

And here's something interesting. Last night I was using CUVIDserver 2006 and DGindexNV 2010 (accidentally, due to MeGUI not using the newest version) and it said 23.976. This morning I used 2010 and 2010, and it said 29.970. So I've spent 15 minutes switching between various combinations (and switching between input files: the 100mb sample, the full m2ts, and the demuxed vc1) and in fact I'm not seeing a pattern at all - sometimes I get 23.976, sometimes 29.970. I've managed to get a DGI file of each, and the one created at 29.970 plays back at the correct length in an AVS file.

So, besides the obvious "what's going on here?", there's the interesting point that in either case, it counts the same number of coded frames and playback frames. Is this correct? Is eac3to wrong about it having soft pulldown?


Edit:
Alright, I've found the pattern (and btw I'm only using 2010 now, so I don't know how this applies to 2006). If I open one of the m2ts files first, it shows 29.970 and any of the test files I subsequently open (inclding the demuxed vc1 stream) show as 29.970. If I then close and re-open DGindexNV, and open a vc1 first, it detects as 23.976 and opening the m2ts afterwards will make it show as 23.976.

[Guest]

Alright, I've found the pattern (and btw I'm only using 2010 now, so I don't know how this applies to 2006). If I open one of the m2ts files first, it shows 29.970 and any of the test files I subsequently open (inclding the demuxed vc1 stream) show as 29.970. If I then close and re-open DGindexNV, and open a vc1 first, it detects as 23.976 and opening the m2ts afterwards will make it show as 23.976. Thank you for your trouble report. I wish they were all as thorough and well presented. :)

I've duplicated this with your samples. Investigating...

[Guest]

The stream is 29.97 fps without pulldown.

Regarding the initial sample.vc1 you linked: I don't know what you demuxed that with but it really is 23.976. Let's forget that because I don't know how you made it. If EAC3TO did that you should report it. If you demux with DGIndexNV 2010 from your sample.m2ts, the resulting sample.demuxed.vc1 opens as 29.97, as it should.

I've located the bug that you exposed. When you open a new file from within DGIndexNV (without closing and reopening DGIndexNV), there are cases where the new sequence headers are not processed. One case is if the only change is the frame rate, which applies in your scenario because you open sample.vc1 first and then open sample.m2ts.

The fix will be in the next release. Your workaround is to close DGIndexNV before opening new files.

(BTW, I have deleted your full sample link for copyright reasons.)

[RedDwarf1]

I'm interested in purchasing a nVidia card so I can use DG NV Tools after having great difficulty decoding BBC HD video using software methods. I have a HD3850 512MB video card currently where the cards idle power consumption was an important thing when I decided to purchase it. ATI latest cards do use less power than nVidia, probably due to AMD's CPU design experience, it's not a myth like I have seen mentioned. It's a pity that ATI doesn't get it's act together and provide a means to program for it's GPU's like nVidia does.

Seeing as there is no ATI GPU assisted decoder for AVISynth and there seems like very little chance of one being produced, I have decided to add a nVidia card as a 2nd Video card in my PC and the card needs to have very low idle power consumption but provide good decoding capabilities. I have read through most of this thread and a large part of the NV Tools development notes and nVidia correspondence and seen people asking about the decoding performance of various cards and seen no definitive guide to allow people to decide which card to purchase. Really doing such tests requires the same PC setup and multiple nVidia cards which people do not usually have access to. nVidia are really the people who could provide such information but it's whether their figures could be trusted.

Does anyone know of any figures to indicate the performance of various nVidia cards, especially when de-interlacing 1080i content?

What type of performance could I expect from a GT 240 DDR5 512MB/1GB card? I would be particularly interested in decoding figures from such a card.


But any performance gain is secondary to the other advantages, also described several times: reliable random access, GPU deinterlacing, GPU resizing/cropping, correct decoding of AVC and VC1, consistent integrated solution to AVC/MPEG/VC1.
Random access and GPU De-Interlacing is important to me providing the quality is good. Are there any examples of the De-Interlace quality?
GPU Resizing would also be of interest providing the quality was good. How does it compare to Spline36resize or Spline64resize?

neuron2, I'm thinking of donating for your DG NV Tools and I have some questions I hope you can help me with.

The Purevideo deinterlacing function looks great, and that's what I guess I will be using most. The thing is my work computer is also my gaming computer. I'm currently using a Radeon 4870 and need an nVidia card. I have come up with 2 choices

1) Get a Fermi (GTX 470 is my first choice)

2) Continue using my Radeon 4870 and get a cheap GT210/GT220 and use it alongside with the 4870 purely for Purevideo.

My question is, is the purevideo deinterlacing quality dependent on the card or are there differences? Also, do you know if your software will work in scenario 2? I'm aware that only Windows 7 supports heterogeneous video card support. I'm currently on vista and will have to upgrade to 7 for that, but I hope to find out whether it will work first.

Thanks!
You are in a similar position to me, except Gaming is of no interest to me. The later last generation nVidia cards are better than earlier cards in a number of ways. The decoding engine, VP4 is much improved on GT210/220/240 cards supporting improved resizers and eliminating the restrictions on H264 decoding resolution sizes that previous cards had. Their idle power usage is very low which is of benefit when not using the cards for decoding. I might of been interested in the GT430 which was meant to be released fairly soon but I don't want to have to wait any longer. So the GT 240 DDR5 is the card that interests me due to it's higher memory bandwidth and low power usage of about 6.3 Watts. Having single slot designs available also helps in my already overcrowded case so I do not lose any other slots.

Would I experience any difference between 512MB and 1GB of onboard memory? How would 1GB benefit me over a 512MB card?

Are there any restrictions on the number of H264 streams that can be opened at the same time? Can it open 2 or 3 different streams and process all at the same time in one AVISynth script?

http://en.wikipedia.org/wiki/Purevideo#The_Fourth_Generation_PureVideo_HD
nVidia Card Feature table (http://www.techarp.com/showarticle.aspx?artno=88&pgno=8)
Video card Power Usage (http://www.xbitlabs.com/articles/video/display/gpu-power-consumption-2010_3.html)

[Guest]

Random access and GPU De-Interlacing is important to me providing the quality is good. Are there any examples of the De-Interlace quality? There is at least one example in this thread.

GPU Resizing would also be of interest providing the quality was good. How does it compare to Spline36resize or Spline64resize? Again, if you read the thread you'll find that CUVID supports only bilinear at this time.

Would I experience any difference between 512MB and 1GB of onboard memory? How would 1GB benefit me over a 512MB card? As described in this thread, the number of simultaneous instances of the decode that you can create is limited by the card's memory.

Are there any restrictions on the number of H264 streams that can be opened at the same time? Can it open 2 or 3 different streams and process all at the same time in one AVISynth script? It depends on the card's memory.

[paulvdb]

Version 2010 now successfully avoids treating Teletext as audio :-)
Unfortunately it still seems to detect BBC HD ac3 2/0 as 3/2. But no biggie.
That is probably the BBC's fault. They sometimes mess up their audio and flag it as 3/2 even though only the left and right channel actually have audio.

[Clumpco]

That is probably the BBC's fault. They sometimes mess up their audio and flag it as 3/2 even though only the left and right channel actually have audio.
If you read my later reply it was the .ts that was corrupted.
But I do know about their 3/2 lockdowns (we'll probably get them again for the duration of the world cup) but in the case in question it was really 2/0.

[WasF]

@RedDwarf1 regarding GPU processing quality, and for who ever asking himself the same questions.
I can only give my subjective impressions, but I've done quiet a bit of video encoding since 2002..
* Resizing: DG NV tools are aimed mainly at HD content. In that case, resizing, if any, means downscaling. Bilinear will do a great job here. Actually, when it comes to downscaling, almost every algo is a champ. I am very satisfied with what CUDA does here. And bilinear is quiet good at upscaling too, although I did it only once with CUDA.
* De-interlacing: when the source is clean, I get great de-interlacing with DG NV. However. I'm encoding a TV show DVD right now (NTSC), and was not satisfied with what CUDA does here. But it's comparable to what I get from a "professional" software player (TotalMedia Theatre 3, "hardware acceleration" disable), so, can I really blame CUDA?!
What I know, is that I get a better looking image with VLC de-interlacing (blend), and an amazing result in AviSynth with eedi2/tdeint (I can post examples if you want, but basically, all lines are straight and smooth with no "stairs" effect with eeid2/tdeint). Of course, I pay dearly for this quality on the CPU level.. (2.65 fps is brutal, but my x264 options are pretty wild, too :))
* Cropping: don't buy CUDA for this! Cropping is so cheap, it almost makes no difference if you do in CPU or GPU.. But if you can make your GPU do it, why the heck not?!
* Memory: this is the topic that brought me to this forum. I have a 256MB 8600 GT (I'm anti-gamer, is you wish), and that's worth 1 1080p stream.
I can do up to 3 720x540 standard def concurrent instances, barely.
This doesn't mean these numbers scale up linearly with memory quantity, though.. it just gives a hint: if you go for it, go heavy on the DDR!
Other testimonials would definitely shed some more light on this specific topic

Cheers.
WasF/

[Guest]

* Cropping: don't buy CUDA for this! Cropping is so cheap, it almost makes no difference if you do in CPU or GPU.. But if you can make your GPU do it, why the heck not?! You miss the point. The most common process is to crop and then rescale. So if you rescale on the GPU, you *must* therefore crop on the GPU.

[WasF]

Of course, neuron2 :).. some features go hand in hand, indeed.
Still, you can do cropping alone on the GPU.

And I don't see where the obligation to GPU-crop comes from if I do resizing: I've been doing DGMultiSource(resize_w=720, resize_h=540) for 6 episodes now, with no cropping, and it works just fine (DGI: CLIP 0 0 0 0) and I don't see what would be illegal if I added an Crop() to my AviSynth script on top of that, too..
You got me worried now!
If you'd care to elaborate, I'd be curious to know what I'm missing..

[woah!]

ok so if you crop after you have used (resize_w=720, resize_h=540) then it wont be 720x540 afterwards would it... it will be smaller, so cropping first and then resize will give you 720x540.

[Guest]

The calculations are more straightforward if you crop first. And if you crop first, then you must do it on the GPU if you do the resizing on the GPU.

As I said, you missed the point. :)

[WasF]

Oh! I got it, I got it!
Yes, if you do GPU resizing and need cropping, it would be much easier and safer to do it all in GPU.
I was more thinking technical limitation or somethin' .. :eek:

Believe it or not, the first time I used DG NV tools, it was for re-encoding Avatar, and I did resizing with CUDA and a bit of cropping in the AVS script :rolleyes:
('Never understand the 8 Mbs/s thing! What was H.264 for, again?!!)