New Version of Avi.net

[audioman]

My movie is 100 mn long
I choose 640 most of the time

[ricardo.santos]

if it was me and it was a 2:35 movie i would choose between 1000 and 1100 no more than that.

[iNFO-DVD]

This is pretty weird, How can a 896 Mo movie give a very good quality ? I am quite surprisedWell you shouldn't be........ Some widescreen movies encoded to 700MB can look perfect, all depends on the movie.

[audioman]

Okay

I have remarked something

When the movie is loaded into avi.net , the QF starting point is approximatively 0.20 , with a given bitrate

In my case , I always keep AC3 in my movies
Obviously when I tick the AC3 button , QF and bitrate get lower What I do is :

increase the file size in order to match the original bitrate

Then I run the comptest and encoding

Is it the right way to proceed ?

I am testing on a short clip right now

By the way , Hi INFO :)

[iNFO-DVD]

Well, when changing codec parameters the compressibility will change.
Exactly, and I can use this to my advantage for extremely low/high compressible movies, hence counter measures.

If you're going to repeat the comptest with the new settings, remember there are few other "counter measures" to add (like lowering resolution or heavy filtering).
I agree but I wouldn't do yet another comptest after changing settings, I feel it would be an overkill. When the comptest % goes out of some set thresholds I'll just have to make sure the internal codec/filter changes I make are set specifically for them thresholds.

IMHO lowering the resolution is the way to go so you can mantain a good encoding speed (even with slower codec settings) and a good overall quality.
I agree.

a) Inform the user that quality is too low and avi.net is reducing the res
b) Keep a SAP compatible resolution
c) Offer an option in the program settings to override the resize (and go on with the user selected resolution) or an option to disable the comptest.
a) If I do that I surely will.
b) Most certainly.
c) Definately.

Thanks for your feedback and I completely agree with you. :thanks:

[iNFO-DVD]

When the movie is loaded into avi.net , the QF starting point is approximatively 0.20 , with a given bitrate, in my case , I always keep AC3 in my movies. Obviously when I tick the AC3 button , QF and bitrate get lower
Yes, rather than you load a movie in and everything say zero or whatever I've always thought it's better to at least get a starting point. To be honest, you could always leave it on what was recommended and a lot of the time the results would be fine. Mainly, it's a starting point, a guess. As for you changing the AC3, yes, the bitrate and QF will lower, and if we're talking about a full movie for example then I'd then probably go to the next set size up and check the QF value to make sure it hasn't went too high. But in saying all that, the comptest will ease your mind that bit more.

Then I run the comptest and encoding, Is it the right way to proceed ? I am testing on a short clip right nowProblem is, though better than nothing, testing on a clip you may have taken from the movie may not give you as acurate results as letting avi.NET do the comptest on the whole movie at set intervals. I'd just set everything up and let it start, if something was wrong, like the value far too high/low then I'd just stop it, change some settings, and go again.

The time it would take to press stop, reload the movie in, choose your size and go again is going to be far quicker than taking a clip then loading that in and doing just the same on it, you'd then have to stop and load the main movie back in again anyway. More often than not your comptest value I would guess would be OK and over time you'll get to know a lot more.

By the way , Hi INFO :)Hi :)

[audioman]

Problem is, though better than nothing, testing on a clip you may have taken from the movie may not give you as acurate results as letting avi.NET do the comptest on the whole movie at set intervals. I'd just set everything up and let it start, if something was wrong, like the value far too high/low then I'd just stop it, change some settings, and go again.
Hi :)

:thanks:

Is 5% a sufficient value to detect high action scenes in a film or non action scenes ?

[iNFO-DVD]

Is 5% a sufficient value to detect high action scenes in a film or non action scenes ?Well, I suppose it's a good enough realistic value, 100% would be best but that's just going to far, a big overkill IMO. 5% seems to be about the accepted 'realistic' value to use. It's doing the 5% at a rate of 14 frames every 280 throughout the whole movie so it is getting good representation of all the scenes.

[MarkCurly]

...I agree but I wouldn't do yet another comptest after changing settings, I feel it would be an overkill. When the comptest % goes out of some set thresholds I'll just have to make sure the internal codec/filter changes I make are set specifically for them thresholds.
It all sounds like it's coming together... :)

1) If the comptest is run, and an initial %'age is calculated, and THEN the user changes the target encode parameters, which 'profile' is then used?
Does avi.NET allow a new 'estimated' comptest value to be calculated based on the new target parameters and change the profile used? (It would be good if it did.)

An example, suppose I chose 1 CD encode (700MB), and the comptest said it was 35% (and would then use your LQ profile).
Now that I know the source is not very compressible, I change the target to a 2CD (1400MB) and then proceed to encode.
It would be good if avi.NET figured out that with the new target size, the estimated comptest is NOW ~70%, and it should now use the HQ profile.
It would not be too good if it still used the LQ profile, based on the old comptest!

2) Does the comptest encode currently compress with the source framesize, or does it use the target framesize?
I know I suggested the comptest encode might use the target framesize, BUT if the comptest encode used the original source framesize and in fact used NONE of the target encode parameters, then the calculated "100%" comptest value is a CONSTANT for a particular source.
Then if the user changed the target parameters, a new comptest value can be calculated EXACTLY based on the new target Qf, just calculate it as Qf(target) / Qf(Source) and the divisor does not have to be re-calculated.

(Of course even if the initial comptest encode *does* use the target framesize, the you can still calculate a new comptest %'age when the target size changes, it will just be a little less accurate...)

3) Does the comptest work for DivX?
If not, chasing some of the links you posted shows that some people use VDubMod as part of a comptest because it can be made to output a framesize file. If you used that file rather than the XviD pass file, it could work for DivX also.

4) Can I change the b-Frames to max consecutive 2 somewhere?

5) A rhetorical/philisophical question about the settings used for each profile...
When deciding what parameters to use for each value of the (MeGUI?) comptest %'ages (profiles), I imagine that the XviD community spend man months or more tuning the matrices and other encode parameters and re-encoding video and viewing it and testing it with video quality programs etc.

In converting those profiles for avi.NET's version of those profiles, it *may* be that some of the parameters are not optimal, because it is not an apples-with-apples situation (avi.NET doesn't want to use custom matrices). Or the threshholds might not be in best place; maybe the UHQ profile should kick in at 80% rather than 90%? Maybe the threshholds would be better if they were all shifteding by 10% in some direction?
(Note: I have NO IDEA if the profiles are not optimal, and I'm not suggesting that the UHQ threshold needs changing, or that all the threshholds need shifting, I'm just using those as examples. (Did you read my earlier post where I said I had no idea what I'm talking about ;)))

Hopefully people can provide feedback on how well they work.

THANK YOU THANK YOU for the wonderful encoding tool that is avi.NET!

[MarkCurly]

Current Settings: (opinions sought and welcomed)

05/09/06 06:49 BETA 2

XX-39 [PROFILE LQ] H263 | 1,1.62,0 | 10,15,20 | 20,10 | 2,15
40-59 [PROFILE GQ] H263 | 1,1.62,0 | 05,10,15 | 10,05 | 2,09
60-89 [PROFILE HQ] MPEG | 1,1.62,0 | 05,10,15 | 05,05 | 2,07(I6)
90-XX [PROFILE UHQ] MPEG | 1,1.00,1 | 00,05,05 | 00,00 | 1,05(I4)

KEY: COMPTEST|PROFILE|QUANT|B-FRAMES|OVERFLOW|COMP CURVE|QUANTIZERS
Have you seen the ESS and ETK "Unofficial" profiles that henryho_hk created (in one of the links you posted)?
http://forum.doom9.org/showthread.php?p=850423#post850423

As near as I can tell, henryho_hk has built a BAT script that automates encoding using an xvid_encraw program.

His script uses the 'normal' (non-SAP) profiles that Teegedeck suggested BUT ALSO has some of his own "unofficial" SAP profiles.
Look in the CODE block at the link above (click+drag mouse to select it all), and can see he has some ESS and MTK profiles.
They are:
comptest >30% ESS
comptest >45% ESS
comptest >58% ESS
comptest >90% ESS
comptest >30% MTK
comptest >45% MTK
comptest >58% MTK
comptest >90% MTK

I don't know if they are any good, or better than yours, but there they are...

Looking at his profiles, there are some things I'm not too sure about...
- "-nopacked" my SAP handles packed OK.
- "-imin 3 -imax 5" the quantizer restrictions seem rather narrow (this for the cmprgt30ess profile, but the others also have narrow quantizer restrictions).
- "-max_bframes 1" my player handles 2 OK
- They don't use the built-in XviD profiles...

(PS: My player is a Philips DVP642)

[iNFO-DVD]

1) If the comptest is run, and an initial %'age is calculated, and THEN the user changes the target encode parameters, which 'profile' is then used?
Does avi.NET allow a new 'estimated' comptest value to be calculated based on the new target parameters and change the profile used? (It would be good if it did.)

An example, suppose I chose 1 CD encode (700MB), and the comptest said it was 35% (and would then use your LQ profile).
Now that I know the source is not very compressible, I change the target to a 2CD (1400MB) and then proceed to encode.
It would be good if avi.NET figured out that with the new target size, the estimated comptest is NOW ~70%, and it should now use the HQ profile.
It would not be too good if it still used the LQ profile, based on the old comptest!
Unfortunately, 700MB at 35% doesn't then equal 70% if set to 1400MB, it would be far more % than that.

I've probably called the profiles with the wrong name, it could be confusing, making a user think the LQ would mean it will make a low quality encode e.t.c, it doesn't mean that. What it means is more like a low % has been detected so I have to use some high encode settings to try to obtain as much quality as I can get from the codec because the user has set it too low. For instant, a LQ encode, especially with DIVX could end up taking far longer than any other as I have to then change the encode/performance setting, turn on some psychovisual and enchanced texture to try to improve it, options I wouldn't normally need to use if there was more bitrate the movie had to play with.

If a % is too low or too high I have to take counter measures to try and improve the quality if too low or make a high % lower by using some sharper quant type for example. I may have to include some filters too, dunno yet, maybe a smoother resizer for low % detections.

2) Does the comptest encode currently compress with the source framesize, or does it use the target framesize?
The source, thanks to you that is. As of the BETA 2 it was switched to the SOURCE as I was actually using the encode size to start with. When I release BETA 3 at some point today that has been changed again slightly to the size of the actual input without borders, silly me was using the direct input size, 720x576 for example rather than the processed size without the black borders, example 720x400 or whatever it may be. It's a lot more acurate now.

Then if the user changed the target parameters, a new comptest value can be calculated EXACTLY based on the new target Qf, just calculate it as Qf(target) / Qf(Source) and the divisor does not have to be re-calculated.
You have a really good point there, hadn't thought of that, infact, I'm really glad you mentioned it. Trouble is, I can't stop the encode to allow user intervention then carry on kind of automatically, that would be unusable in reality, whatever I do I have to make that decision from within the program automatically. At the very least I could display on the screen a % estimate of every other set size in the list.

3) Does the comptest work for DivX?
Yes, it's all done, in BETA 3. I noticed the DIVX % are lower than the XVID ones, sometimes quite massively, I verified this behaviour with AutoGK too.

4) Can I change the b-Frames to max consecutive 2 somewhere?No, 2-bframes are NOT fully SAP compliant.

5) A rhetorical/philisophical question about the settings used for each profile...
When deciding what parameters to use for each value of the (MeGUI?) comptest %'ages (profiles), I imagine that the XviD community spend man months or more tuning the matrices and other encode parameters and re-encoding video and viewing it and testing it with video quality programs etc.

In converting those profiles for avi.NET's version of those profiles, it *may* be that some of the parameters are not optimal, because it is not an apples-with-apples situation (avi.NET doesn't want to use custom matrices). Or the threshholds might not be in best place; maybe the UHQ profile should kick in at 80% rather than 90%? Maybe the threshholds would be better if they were all shifteding by 10% in some direction?
That's just going to be trial and error I think, all I can do is read as much as possible about the codec settings and try to chose what's best within the threshold, not to mention the constant clip encodes I've been doing. I'm open to suggestions on this front. As for thresholds, I have set the DIVX thresholds slightly lower that the XVID ones.

I'm eagar to hear suggestions on particular codec settings, both XVID and DIVX, for the different profile thresholds.

[MarkCurly]

One last thing.... Does anyone use the 352 screen size option? Shall I drop it and use something between 480 and 640, like 576?
560 would be closer to the geometric mean of 640 and 480.
I would have thought that smaller widths (and that includes 480) would be less often used.
How about adding two imtermediate sizes between 640 and 480, like so: 640, 576, 528 and 480?

[iNFO-DVD]

Have you seen the ESS and ETK "Unofficial" profiles that henryho_hk created (in one of the links you posted)?
http://forum.doom9.org/showthread.php?p=850423#post850423No, I'd missed them, cheers, printing them out now, will take a look at them with a coffee.

I must admit, what I don't like is different profiles for different chipsets... what if you had ESS and MTK? I have a SIGMA and MTK chipset, my mate has MTK and ESS, what profile do they use then? What if you got rid of your ESS and got an MTK, oh no, you may not be able to play your AVI's.... I'd always stick with a more global profile.

Anyway, off to look through them profiles, thanks again.

[iNFO-DVD]

560 would be closer to the geometric mean of 640 and 480.Yes of course, that's what I meant.... :D

[MarkCurly]

I've probably called the profiles with the wrong name...
I think they are OK.
I do understand them to mean:
Make the best choices for what is likely a LQ encode.
Make the best choices for what is likely a GQ encode.
Make the best choices for what is likely a HQ encode.
Make the best choices for what is likely a UHQ encode.

Trouble is, I can't stop the encode to allow user intervention then carry on kind of automatically, that would be unusable in reality, whatever I do I have to make that decision from within the program automatically.
NOOO!! :eek:
I was hoping the comptest could be called like the bitrate calculator, and the results played around with before the encode continues.
One of the things I don't like about AutoGK is that if the comptest suggests you are going to get a low quality encode, you can't change your mind unless you abort the encode and start again. If you let AutoGK continue, you can end up with a too-small framesize.

No, 2-bframes are NOT fully SAP compliant.
My ESS based player does handle 2 consecutive OK (I once even tried a file with 5% of 3 consecutive).
Any tweakability you can provide here would be much appreciated.

Unfortunately, 700MB at 35% doesn't then equal 70% if set to 1400MB, it would be far more % than that.
Somewhat more, yes. :)
<pedantic>I didn't want to bother with a calculation, but assuming 100MB of the video was MP3 audio, then the original "1CD" is 600MB of video, and the "2CD" is 1300MB of video, a ration of 2.17, so I would expect a new comptest of ~76% (assuming I have the gist of the calculation).
(BTW: when I use "~", I mean it to mean "Approximately")
</pedantic>

[iNFO-DVD]

I think they are OK.
I do understand them to mean:
Make the best choices for what is likely a LQ encode.
Make the best choices for what is likely a GQ encode.
Make the best choices for what is likely a HQ encode.
Make the best choices for what is likely a UHQ encodeYes.

NOOO!! :eek:
I was hoping the comptest could be called like the bitrate calculator, and the results played around with before the encode continues.lol, well at the moment, once turned on it does it automatically, but I'll see what else I can do during development.

One of the things I don't like about AutoGK is that if the comptest suggests you are going to get a low quality encode, you can't change your mind unless you abort the encode and start again.Just out of curiousity, what does that actually mean in reality refering to avi.NET?

Current:
1. Load movie, do your settings, go.
2. Comptest
3. Is it OK? Most of the time YES, if not, stop, goto 1 and miss 2 and 3
4. Encode

Your way:
1. Load movie, do your settings, go.
2. Comptest
3. Is it OK? if yes, goto 4, if not goto 1 and miss 2 and 3
4. Encode

Doesn't seem to be much difference apart from the fact the comptest everytime would have to be verified and back to user to carry on with encode. It seems the current way would actually be quicker and less hassle because after all most of the time the % value will be acceptable. But saying all that, I know what you mean. :)

My ESS based player does handle 2 consecutive OKI know, but some others don't.

then the original "1CD" is 600MB of video, and the "2CD" is 1300MB of video, a ration of 2.17, so I would expect a new comptest of ~76%No, it doesn't change at the same rate..... in your example I'd guess more round the 105% mark. But it doesn't matter as your idea of the QF was perfect anyway.

[MarkCurly]

Current:
1. Load movie, do your settings, go.
2. Comptest
3. Is it OK? Most of the time YES, if not, stop, goto 1 and miss 2 and 3
4. Encode
Whew! :D The optional stop in step 3 allays most of my fears.

My ideal flowchart goes like so:
1. Load movie.
2. Do settings. Can include calling the bitrate calculator to help with the settings. Can include changing the LQ/GQ/HQ/UHQ profile used (even before a comptest has been done). If a comptest has already been done and a Qf(comptest) value is available, then auto-suggest the profile based on Qf(target)/Qf(comptest).
3. Optional manually called comptest. The profile (LQ/GQ/HQ/UHQ) is set. Stop and return to step 2. (Now Qf(comptest) is known, step 2 can be smarter.)
4. User says "Go, encode"
5. Optional "auto" configured comptest. The profile (LQ/GQ/HQ/UHQ) is set.
6. Is it OK? Most of the time YES, if not, stop, goto 2 7. Encode

A user would be unlikely to do BOTH the manual step 3 comptest AND the optional auto-configured step 5 comptest.
I would likely turn the step 5 auto-comptest off, and just do the step 3 manual comptest.
Other users might NOT do the step 3 manual comptest, but instead configure the step 5 to auto-run.

No, it doesn't change at the same rate..... in your example I'd guess more round the 105% mark. But it doesn't matter as your idea of the QF was perfect anyway.
That has me completely confused... :confused:
Can you please give a worked example?

[Sharktooth]

The xvid/divx difference in comptest is due to the "optimized H.263" quantization.
Sadly xvid and divx should have different thresholds for comptest.

[weaver4]

if it was me and it was a 2:35 movie i would choose between 1000 and 1100 no more than that.

I also use 640 as width. And on the seldom occasion I do two pass I use I 1000 for 9:5 and I use 1200 for 4:3.

[audioman]

Yesterday night I encoded a dreamworks film
Here is the Log

avi.NET v2.1.8.5
Intel(R) Pentium(R) M processor 1400MHz (1399MHz)

IN : C:\AVINET RIPS\SHREK\VTS_01_PGC_01_1.VOB
OUT: C:\AVINET RIPS\SHREK\Shrek 1318.avi

ORIGINAL MPEG FILE
MPEG TYPE : 2
RESOLUTION : 720x576
ASPECT RATIO: 16:9
FIELD ORDER : TFF

CONVERTED AVI FILE
RESOLUTION : 640x352
DURATION : 1:26:23
FPS : 25
CODEC : XVID
PASSES : 2
METHOD : SIZE BASED
SIZE : 1094MB
BITRATE : 1318Kbps
BITS/PIXEL : 0.23
AUDIO : AC3
SUBTITLES : NONE

SELECTED OPTIONS
DE-INTERLACE: NO
DE-GRAINED : NO
HIGHER CROP : NO

COMPRESSION VALUE : 94%

TIME TAKEN [AUDIO]: N/A
TIME TAKEN [VIDEO]: 03:06:03
ENCODE RATE 1/2 : 37 FPS
ENCODE RATE 2/2 : 28 FPS

TIME TAKEN [TOTAL]: 03:10:12

The QF starting point was 0.23 with a 896 MB size


AVI.NET choose the UHQ profile but I got noisy backgrounds in
dark scenes :(

What should I do ? encode at higher rate ?

I don't really think the image quality will be better

:thanks:

[audioman]

Another weird thing I remarked

Look at the PIC

http://pix.nofrag.com/d6/1a/df241e943d5fea53585ba250e882t.jpg (http://pix.nofrag.com/d6/1a/df241e943d5fea53585ba250e882.html)

In upper right corner Look at the Codec used : DIVX :confused:

I always do XVID and the codec used is DIVX

Do you have this kind of issue too Guys ?

[iNFO-DVD]

The QF starting point was 0.23 with a 896 MB size. AVI.NET choose the UHQ profile but I got noisy backgrounds in dark scenes :(

What should I do ? encode at higher rate ?

I don't really think the image quality will be better
Well the settings in that BETA are not yet finalised so I wouldn't worry too much. Why are you encoding at 94%, drop in to 700MB, I know this sounds strange but the picture may even improve.

[audioman]

Well the settings in that BETA are not yet finalised so I wouldn't worry too much.

It's okay That's what I was thinking :)

Why are you encoding at 94%, drop in to 700MB, I know this sounds strange but the picture may even improve.

94% is the value given by the comptest for a 86mn movie
I kept ac3 audio
I don't really understand the question
Could you develop ?

:thanks:

[iNFO-DVD]

Another weird thing I remarked

Look at the PIC

http://pix.nofrag.com/d6/1a/df241e943d5fea53585ba250e882t.jpg (http://pix.nofrag.com/d6/1a/df241e943d5fea53585ba250e882.html)

In upper right corner Look at the Codec used : DIVX :confused:

I always do XVID and the codec used is DIVX

Do you have this kind of issue too Guys ?Well first of all, update your GSpot, there's a new one that's just came out. Yours is even older than the 2 year old version.

As of v2.1.8.0 of avi.NET the MUX program was changed to DivXMux, instead of VirtualDubMod. Whether you have encoded using DIVX or XVID DivXMux will write the FourCC as DIVX, that's why you think you have a DIVX file, because of the FourCC.

Update your GSpot and you'll notice it will display the XVID version at middle/left, for example XviD0041. If the display and/or log says XVID, then it's an XVID, have faith.. :)

[iNFO-DVD]

94% is the value given by the comptest for a 86mn movie.
I kept ac3 audio
I don't really understand the question
Could you develop ?
You have chosen 896MB as the size, the movie compresses very well (94%) so there's probably no need to have it that high. It's that high, counter measures have kicked in, drop the size, convert to 700MB instead.

[audioman]

No I did not convert to 896 MB
896 is the starting point value given by avi.net with a bitrate 1318

I ticked the ac3 button and the values get lower ( normal )
I increased the values to get a bitrate 1318

Avi.net gave a file size of 1094 mo with a bitrate of 1318

The comptest ran and gave a 94 % compression and decided to choose the UHQ profile

Is that clearer ?

[iNFO-DVD]

No I did not convert to 896 MB
896 is the starting point value given by avi.net with a bitrate 1318

I ticked the ac3 button and the values get lower ( normal )
I increased the values to get a bitrate 1318

Avi.net gave a file size of 1094 mo with a bitrate of 1318

The comptest ran and gave a 94 % compression and decided to choose the UHQ profile

Is that clearer ?I know, it's toooooooooooooooooooooooooooooooooooo high :)

Keep at 893, select AC3, everything will drop, leave it... why keep to the bitrate originally given???? keep to set sizes, it makes more sense.

94% is too high.

[audioman]

Ok

The compressibility test shows 71% with the 893 file size and AC3 enabled
The profile HQ is used :)

[iNFO-DVD]

That's better. :)

[audioman]

No it's not better

The quality is just worse
I don't have the quality of the original

[iNFO-DVD]

Do you have a screenshot of what you are refering to? or a sample of the original? Actually, hold back on the sample as not sure of the legality of it if it's a commercial movie.

On a different note, something I noticed in a couple of your earlier posts:

METHOD : SIZE BASED
SIZE : 1094MB
BITRATE : 1318Kbps

I increased the values to get a bitrate 1318
Avi.net gave a file size of 1094 mo with a bitrate of 1318
As it says SIZE BASED that means you must have set the size, noted the bitrate, selected AC3, then fiddled about with the size to get back to the bitrate of 1318.

Why didn't you just click the bitrate box instead?
As I mentioned earlier wouldn't it have been easier to select a set size of 1120 or 1398? Don't think you have to go back to what avi.NET tells you when you load in the file.

EXPLANATION OF SIZE BASED AND BITRATE BASED OPTIONS:
As you set a size, whether a set size or one you insert manually, the bitrate will change when you hit enter. If you change the bitrate directly by clicking the bitrate box it does not change the size. It's the value in the bitrate box that gets sent to the codec. If you want a set bitrate, click the bitrate box and enter value, your LOG will then say BITRATE BASED instead of SIZE BASED.

Personally I never use BITRATE BASED and also choose a set size for movies as they're going to end up on a DVD.

[iNFO-DVD]

I don't have the quality of the originallol, is that what you are expecting? you may be dissapointed, whatever program you use.

[audioman]

lol, is that what you are expecting? you may be dissapointed, whatever program you use.

I don't share your opinion
There are several posts in this forum , talking about xvid matrices ( 6o9, sharktooth etc ...) with guys really happy to have a great quality near original
But I don't want to deal with Megui or other softwares

I don't mind about the encoding time ( I even think that avi.net encodes too fast !!! Kind of surprising compared to others ) :)
I just want a very good backup quality ( near the original )
I don't want washed backgrounds, and pixellisation
I just want to avoid making a backup DVD because the filesize is too high

I think that your software rocks and is full of simplicity .That's what I appreciate about avi.net
But I did not find the way to get High quality backups :(

[iNFO-DVD]

great quality near originalAh yes, that's better, you originally said: I don't have the quality of the original ;)

[iNFO-DVD]

I even think that avi.net encodes too fast !!!lol, you want say that with the next version......

I don't want washed backgrounds, and pixellisationShow me, would like to see. Did it look the same with other software?

But I did not find the way to get High quality backups :(So everything you encode is bad?

[audioman]

lol, you want say that with the next version......


Ok Let's see :)

Check your email
I sent you a short clip
You 'll see by yourself :)

[iNFO-DVD]

You sent me a movie clip that was 598Kbps.............

598!

What are you expecting at that bitrate?

I can tell you now, for a bitrate of 598 I thought that clip was pretty damn good.

[audioman]

The clip is 1 mn long

Have you checked the txt file ? :confused:

I don' think we have the same eyes
Look at the clouds at the beggining , the image is washed out

I am sure you know What I mean especially in dark scenes
You've said it there
http://forum.doom9.org/showthread.php?p=863098#post863098

I would have advices from other guys too :)

[mod]

Can I see the clip too? :rolleyes:

[audioman]

Hi mod

Check your private message for the Link

[iNFO-DVD]

The clip is 1 mn long
Have you checked the txt file ? :confused:Yes I have, that's were it tells me you encoded it at 598...... You mentioned poor quality, at the bitrate that file was encoded at I'm not suprised. What is it that is confusing you? To be honest, you're puzzling me :confused:

Here's 3 screens from the encode, including the starting clouds. This was encoded at 598. In my opinion, for 598 they're quite good.

http://www.dvduk.pwp.blueyonder.co.uk/__img/_forum/3screen.png

I am sure you know What I mean especially in dark scenes
You've said it there
http://forum.doom9.org/showthread.php?p=863098#post863098Well I can't see anything that looks anywhere near as bad as them DIVX shots.

One more thing to note, that avi.NET you are using is a BETA, it's had it's codec settings messed with.... but more importantly, the bitrate was 598!!!!!

[iNFO-DVD]

Just wanted to quickly mention something and ask a question.


Before I first released avi.NET, back when I was messing with interleave settings and getting VirtualDubMod to mux the A/V I realised that using an interleave of 1 vid frame caused me problems with my player, it would only happen if this was true:

XVID AC3 1 vid frame interleave

The problem was like a sort of stutter, only bearly, but enough to notice, it was unacceptable. So after lots of reading it seemed the way to go was 96/96ms for AC3 and since then this has worked well.

Well........ I'd forget about all that and time went on and recently I changed the MUX program to DivXMux, which gives no control over interleave settings and guess what, yes, it interleaves at 1 vid frame. My problem has returned. :mad:

Now what do I do?

What I'm asking is if anyone else has this problem with their player, if so, I'll drop DivXMux and return to VDM. If it's just kind of me, I'll stick with it.

The problem doesn't occur with any DIVX file, or XVID with MP3, just XVID with AC3.

[mod]

Hmm.. never used XviD + ac3, will try if I have time.
A question: how do you make divxmux work with xvid? I'd like to use it instead of AviMux GUI but I often get sync issues. If I remember correctly you are using a "modified" version right? Is it possible to share it? If it isn't could you just post the arguments for a simple xxx.avi + yyy.mp3 or zzz.ac3 mux? Thanks (hey I didn't forget the comp check, simply I'm too busy atm, I'll work on it in the weekend :) )

EDIT: Thanks audioman for the link (well.. the 2nd one :D ). I agree with iNFO-DVD, imo the quality is really good @598kbps (dark colors, fast motion..). Of course it depends on the output device.

[audioman]

thanks guys

Sorry having bothering you with my screenshots

I 'll do some more tests then in order to get the best quality
I must make me a reason that with xvid or divx codec I will never get the same quality as original :)
I don't want to backup my own dvd in dvd either

Why all you guys don't keep AC3 track in your movies
I personally own a dolby digital system and it is very pleasant

By the INFO, concerning screenshots
Which software do u use to capture ?

:)

[iNFO-DVD]

Well it's been a long time coming, sorry about that, it is just because I have had to do test after test, tweak after tweak, of various clips with takes days.

There has been *BIG* changes in this BETA 3 version. Custom Matrices have now been included. I've tested on 2 different players and they work fine but if you have a player which will not play these AVI's from avi.NET in this version (old ESS?), then de-selecting 'Advanced Analysis' will make avi.NET revert back to what it used to be like, with pretty much standard player profile settings.

Clicking the 'Advanced Analysis' option in avi.NET will cause it, when encoding has started, to run a comptest and enable the internal use of custom matrices. I've included 2 matrices so far. EQM_V3ULR and EQM_V3UR (Thanks Sharktooth), would like to include the EQM_V3HR but I need to do some further tests first. The difference in picture quality when using these matrices has impressed me very much.

SOME CURRENT SETTINGS
comptest value of < 60% AND bitrate < 850 will use: EQM_V3ULR
comptest value of > 60% AND bitrate < 850 will use: EQM_V3LR
bitrate > 850 AND < 1550 will use: EQM_V3LR
bitrate > 1550 will use: EQM_V3HR (not yet, currently still EQM_V3LR)

Codec settings also differ slightly depending on comptest e.t.c.
comptest value is now more accurate than BETA 2.

If just changed the resize back to Bicubic instead of Lanczos as Lanczos can be a bit sharp.
Not sure whether to go with:
Bicubic resize or Lanczos resize with RemoveGrain(mode=2)

Please use the BETA 3 and test, thanks.

Download >> HERE (http://www.dvduk.pwp.blueyonder.co.uk/__files/avi.NET_2185_BETA3.zip) <<

Unzip, run, (it's a self-extracting exe file), point it to avi.NET folder when asked, all done.

[MarkCurly]

...I've tested on 2 different players and they work fine ...
One of those wouldn't happen to be a Philips DVP642/630 would it?

[iNFO-DVD]

These are the current BETA 3 codec settings during encode when 'Advanced Analysis' used.

http://www.dvduk.pwp.blueyonder.co.uk/__img/_forum/SETTINGS.PNG

[ricardo.santos]

ive tried beta 3 and according to my tests/comparisons( 6 so far) the output looks better without the matrices/comptest when using EQM_V3ULR

will post some pics later

[iNFO-DVD]

Are you just refering to the ULR?

I may know what you mean, doing my own test, the LR matrix ALWAYS looked better than the ULR, but it wasn't quite that simple. It looked better by comparing frames one by one, which isn't actually how you would normally see it. By the frame by frame compare I could notice more artefacts around objects with ULR, LR was much crisper, but in saying that ULR isn't supposed to be sharp, it's much better when you are going to such a low stage in bitrate that when blocks start to appear it smoothes them much better, the actual textures in the background e.t.c. Well, that's what my tests revealed. I may have to change the bitrate thresholds slightly.

Strange you say it was worse than without though. Every single one of my tests over this past week, and I mean hundreds, has revealed the matrix tests were ALWAYS better looking, IMO.

[iNFO-DVD]

when using EQM_V3ULR

Was this the type of problem you were getting with ULR?

http://www.dvduk.pwp.blueyonder.co.uk/__img/_forum/EQM_V3ULR.png

Look around the guys face edge, eyes, nose, mouth and the edges of the fingers on the girls hand. The ULR seems to not like edges at all. Using the LR on the above clips comes out fine, better than original.

I need to try and find a low bitrate matrix that at least works with most players, or a lot of them, unless I just stick to the LR for everything under about 1550 bitrate?