AVI.Net problem

[weaver4]

I have used avi.net for a couple of years without a problem, but on the latest version I tried to backup Farscape. I was using XviD. When I try to run the resulting avi in Media Player Classic I get the following message. In media player it simply says the codec cannot be found. It will not play on my either my Philips or Pioneer SAPs. I tried the same movie with Dr Divx 2.0 and AGK without a problem.

Any ideas? Is this a Avi.net bug?

==============================================

Media Player Classic could not render some of the pins in the graph, you may not have the needed codecs or filters installed on the system.


Media Type 0:
--------------------------
Unknown

AM_MEDIA_TYPE:
majortype: MEDIATYPE_Stream {E436EB83-524F-11CE-9F53-0020AF0BA770}
subtype: MEDIASUBTYPE_Avi {E436EB88-524F-11CE-9F53-0020AF0BA770}
formattype: TIME_FORMAT_NONE {00000000-0000-0000-0000-000000000000}
bFixedSizeSamples: 1
bTemporalCompression: 0
lSampleSize: 1
cbFormat: 0


Media Type 1:
--------------------------
Unknown

AM_MEDIA_TYPE:
majortype: MEDIATYPE_Stream {E436EB83-524F-11CE-9F53-0020AF0BA770}
subtype: TIME_FORMAT_NONE {00000000-0000-0000-0000-000000000000}
formattype: TIME_FORMAT_NONE {00000000-0000-0000-0000-000000000000}
bFixedSizeSamples: 1
bTemporalCompression: 0
lSampleSize: 1
cbFormat: 0

[iNFO-DVD]

Well this is a new one. What XVID are you using? What size is the converted AVI? What settings did you use or can you send me the log file?

[weaver4]

I am using Xvid Version 1.1.0. The converted file size is 2.2Gig. Here is the log file.

===========================================
avi.NET v2.1.7.0
AMD Athlon(TM) XP 1800+ (1526MHz)

IN : E:\Movies\Ripped1\Movie\VTS_01_1.VOB
OUT: E:\Movies\Farscape.xan3.avi

ORIGINAL MPEG FILE
MPEG TYPE : 2
RESOLUTION : 720x480
ASPECT RATIO: 16:9
FIELD ORDER : TFF

CONVERTED AVI FILE
RESOLUTION : 640x368
DURATION : 3:03:24
FPS : 23.976
CODEC : XVID
PASSES : 1
METHOD : QUALITY BASED
QUANT : 3.0
AUDIO : MP3
SUBTITLES : NONE

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

TIME TAKEN [AUDIO]: 00:30:23
TIME TAKEN [VIDEO]: 04:16:05
ENCODE RATE 1/1 : 26 FPS

TIME TAKEN [TOTAL]: 04:49:05

[iNFO-DVD]

That's the problem, you've gone over 2GB.

[audioman]

Hi Weaver

Why do you choose single pass QUANT 3 instead of choosing a size based movie

Is this influence the final quality of your movie ?

Thanks for your answers

[weaver4]

The quality of single pass is every bit as good as the quality for two pass for the same filesize. The problem with single pass is that you don't know what the final size will be; all you know is that you will have the quality you want when it is complete.

The problem with two pass is that no one can guess at what the final size of the movie needs to be to get the quality that you want. For example the Movie Firewall will compress to 700k very nicely but the movie Casablanca takes 1.7g. Getting a good quality picture for a given bitrate is so subjective I don't think it can be done reliably. It depends on how much "black" is in the movie, how much action, how busy are the backgrounds, the resolution of the movie, and several other factors.

So, I just set the quanitizer to 3 (XviD) for movies that I want a good copy of and 2.5 for those movies I want an excellent copy of; and I am done with it.

[Sharktooth]

That's why a compression test is needed...

[iNFO-DVD]

The only use of a compression test in avi.NET would be if you wanted to do a single pass quant encode but to a specific size.

[Sharktooth]

also to predict quality->file size.
assuming q2 is the best possible quality (and bigger filesize), with a small comptest, quality can be predicted for nth pass modes too.

[audioman]

The quality of single pass is every bit as good as the quality for two pass for the same filesize. The problem with single pass is that you don't know what the final size will be; all you know is that you will have the quality you want when it is complete.

The problem with two pass is that no one can guess at what the final size of the movie needs to be to get the quality that you want. For example the Movie Firewall will compress to 700k very nicely but the movie Casablanca takes 1.7g. Getting a good quality picture for a given bitrate is so subjective I don't think it can be done reliably. It depends on how much "black" is in the movie, how much action, how busy are the backgrounds, the resolution of the movie, and several other factors.

So, I just set the quanitizer to 3 (XviD) for movies that I want a good copy of and 2.5 for those movies I want an excellent copy of; and I am done with it.

This is a very good explanation Weaver thanks a lot

I personally own a tvix and I really do not care about file size and I want a very good copy of my movies
( I just do not want to get a 4 gig avi movie file or else in that case it is no use encoding DVD to avi .... Better to make a dvd copy of the original movie )

Do you understand ?
Do you often get 4 go avi movie files ?

Thanks

[iNFO-DVD]

also to predict quality->file size.
assuming q2 is the best possible quality (and bigger filesize), with a small comptest, quality can be predicted for nth pass modes too.To what goal? avi.NET is not AutoGK, I'm not going to do a comptest then with the results start changing resolutions, b-frames, quant thresholds or using different matrixes, I have standalone compatibility to worry about, top priority, I can only change certain things. Let's say I done a comptest, what would I then do with the results? Stop the encoding process and ask the user to choose a larger resolution because actually this movie compresses quite well, or vice versa, is that realistic? I can't see users being very happy about that.

After setting your desired output resolution and required target size, if after that a compression test would help, then I'd like to know how, if it would benefit the quality of the output file with what I have to work with then I'm definately interested.

[ricardo.santos]

I´m not an expert but im going to give my opinion about this comptest requests:

1- if you are not sure your choosen bitarate/size is not good, you can always convert the 5 or 10 minutes of the movie(dvd shrink in reauthor mode) before you convert your main movie. For example you have a 120 minute film that youre not sure will look good on 1200 megs, extract 10 minutes of the movie and set desired size as 120 megs, at the end you´ll have (i think) a good idea of what the movie will loook. No need for extra coding or requests. just do quick test yourself before your conversion.

if Info-dvd enables a "comptest" on avi.Net i think it would be a nice idea to leave it up to the user to decide to use it or not.

My 2 cents

[audioman]

Weaver

Can you answer my question please

Thanks a lot :)

[ricardo.santos]

I personally own a tvix and I really do not care about file size

why are you encoding your movies to divx, if you have no file size problems just make a backup of your dvd.

[ricardo.santos]

The quality of single pass is every bit as good as the quality for two pass for the same filesize

???!!!????

are you saying the quality of a 1400 single pass conversion is the same as a 1400 2 pass conversion?

and then you say:
The problem with single pass is that you don't know what the final size will be

???!!!!

[unskinnyboy]

To what goal? avi.NET is not AutoGK, I'm not going to do a comptest then with the results start changing resolutions, b-frames, quant thresholds or using different matrixes, I have standalone compatibility to worry about, top priority, I can only change certain things. Let's say I done a comptest, what would I then do with the results? Stop the encoding process and ask the user to choose a larger resolution because actually this movie compresses quite well, or vice versa, is that realistic? I can't see users being very happy about that.

After setting your desired output resolution and required target size, if after that a compression test would help, then I'd like to know how, if it would benefit the quality of the output file with what I have to work with then I'm definately interested.
Standalone compatibility won't be affected by changing resolutions (unless it is > 720x???), b-frames (unless you go above 2) or quantizers (bitrate must be limited using VBV, but I reckon you already do that). Matrices will affect compatibility, so when swapping matrices based on the comptest results, you must limit yourself to matrices which are tested and proven to be standalone compatible - HVS series, EQM series etc.

You can't be asking the users to intervene in between because it will be annoying for them to sit and wait for the comptest results everytime. Any adjustments to the XviD settings or the avs script must be made internally, transparent to the end user. But then avi.NET runs the risk of looking like an AutoGK clone.
I´m not an expert but im going to give my opinion about this comptest requests:

1- if you are not sure your choosen bitarate/size is not good, you can always convert the 5 or 10 minutes of the movie(dvd shrink in reauthor mode) before you convert your main movie. For example you have a 120 minute film that youre not sure will look good on 1200 megs, extract 10 minutes of the movie and set desired size as 120 megs, at the end you´ll have (i think) a good idea of what the movie will loook. No need for extra coding or requests. just do quick test yourself before your conversion.

if Info-dvd enables a "comptest" on avi.Net i think it would be a nice idea to leave it up to the user to decide to use it or not.

My 2 cents
Won't work. Compressing a 10 minute section of the movie won't give you an indication of how the movie will compress as a *whole*, unless the movie ambience look the same throughout the whole movie which is seldom the case for your typical movie. The strength of the comptest lies in the fact that you are covering the whole movie length by sampling the movie at regular intervals using the SelectRangeEvery() Avisynth function. Thus you are taking into account the compressibility of the movie as a whole.

[iNFO-DVD]

Standalone compatibility won't be affected by changing resolutions (unless it is > 720x???), b-frames (unless you go above 2) or quantizers (bitrate must be limited using VBV, but I reckon you already do that).Yes. As for the resolution I prefer to have a set resolution and work around that whereas AGK will change it if need be if the default 'Auto' setting is used. Just seems strange to me to do it like that but everyone has their own opinion on that and will obviously use whichever program suits their needs. As for the B-Frames, I play safe with just 1. My standalone player and my friends player (different chipset) do not like 2 B-Frames.

tested and proven to be standalone compatible - HVS series, EQM series etc.HVS and EQM, hmm, I might look into that, can you asure me these are pretty much 100% standalone compatible? I'll have to look deeper into that as using these may be a viable option if I can see any improvement in quality.

But then avi.NET runs the risk of looking like an AutoGK clone.Exactly...... :goodpost:

[unskinnyboy]

HVS and EQM, hmm, I might look into that, can you asure me these are pretty much 100% standalone compatible? I'll have to look deeper into that as using these may be a viable option if I can see any improvement in quality.
I can only guarantee this much - that I have never heard anyone having issues with the above matrices, while I have heard people complain about others, namely Jawor's 1CD & 2CD. Now don't shoot me if some standalone on some corner of the earth refuses to play these matrices. Actually, most of the newer standalones shouldn't be having any issues playing anything you throw at them, except for 3-warppoint GMC. manono often mentions Oppo players (http://www.oppodigital.com/) as rather versatile. But I also understand that you are aiming for maximum compatibility which is why I hesitate to vouch for anything.

Only one way to know for sure - implement the feature and wait for user feedback.

[iNFO-DVD]

OK, thanks, I'll do some tests and try them on my mates player, that chokes on most things.

[weaver4]

Weaver

Can you answer my question please

Thanks a lot :)

I am not sure I understand your question. But I will try.

I find that I get about a 1/4 to 1/5 compression when I use Xvid Q 3 or DivX Q 4. If I set the quantizer to 2.5 for Xvid or 3.5 for DivX I get compression of 1/3 to 1/4.

So sometimes I can get 4 movies on a DVD and sometimes only 3; depending on the length of a movie.

[weaver4]

???!!!????

are you saying the quality of a 1400 single pass conversion is the same as a 1400 2 pass conversion?

and then you say:


???!!!!

No what I am saying is that if you compress a movie using Xvid quantizer of 3 and the movie comes out to be 900kB long (for example). Then you go back and run a two-pass, setting the file size to 900kB. Then if you compare the single-pass movie and the two-pass movie you will not see a quality difference.

With a single pass you simply set the quantizer to a value and then execute. You have no idea how big the file will be until the encoding is complete.

[audioman]

So guys , what kind of setting would you advise me to use

I tested the weaver method and the final avi file is pretty good according to my eyes

[audioman]

I can only guarantee this much - that I have never heard anyone having issues with the above matrices, while I have heard people complain about others, namely Jawor's 1CD & 2CD. Now don't shoot me if some standalone on some corner of the earth refuses to play these matrices. Actually, most of the newer standalones shouldn't be having any issues playing anything you throw at them, except for 3-warppoint GMC. manono often mentions Oppo players (http://www.oppodigital.com/) as rather versatile. But I also understand that you are aiming for maximum compatibility which is why I hesitate to vouch for anything.

Only one way to know for sure - implement the feature and wait for user feedback.


It is pretty good :goodpost:
I am looking forward your tests INFO :D
testing those kind of settings on my TVIX is gonna be great

[ricardo.santos]

After a couple of tests:

10 minutes vob: 346 MG (extras from stargate DVD where theres quiet moments, "action" moments and CG moments)

divx QT4: 106 MG
Divx QT3: 148 MG
Couldnt see any difference between the 2 both converted with avi.Net under 10 minutes each.


xvid QT3: 112 MG
xvid QT2,5: 152 MG

there were differences between xvid QT3/2,5 ; XvidQT3 looked worse than DivxQT4, both converted with avi.Net

Divx QT4 looks the same as Xvid 3/2,5

Xvid encodings took 16 minutes and a few seconds each

divx Qt4 looks to me the best one.

[Sharktooth]

OK, thanks, I'll do some tests and try them on my mates player, that chokes on most things.
info on EQM V3 matrices: http://forum.doom9.org/showthread.php?s=&threadid=83125

There is a note on matrices not compatible with "picky" standalones.
EQM V3-HR can be modified to be compatible just replacing all 15s with 16s (in inter matrix).
AutoGK just uses EQM V2 as a sharp matrix (link in my sig).

[iNFO-DVD]

info on EQM V3 matrices: http://forum.doom9.org/showthread.php?s=&threadid=83125

There is a note on matrices not compatible with "picky" standalones.
EQM V3-HR can be modified to be compatible just replacing all 15s with 16s (in inter matrix).
AutoGK just uses EQM V2 as a sharp matrix (link in my sig).OK thanks, will do some reading.

[iNFO-DVD]

I tested the weaver method and the final avi file is pretty good according to my eyesThe 'weaver method', I love it, lol, almost sounds official. :D

[weaver4]

info-dvd. Next time you make some changes to avi.net could you make a very small one for me. Could you have it remember the options across sessions? It does not remember quantizer settings now.

Since you are using .Net I have a class that will manage a ini file if you would like like a copy of it.

[ricardo.santos]

Hi everyone

been testing 1 pass quality modes/need for comptest since this thread "heated up".

can anyone tell me how i can make a comptest using the SELECTEVERYRANGE avisynth filter?

i found this but as im not that good with avisynth cananyone fill in the gaps

This filter is available starting from v2.5. The filter selects length number of frames every n frames, starting from frame offset. Default values are: Select 50 frames every 1500 frames, starting with first selection at frame 0. (every = 1500, length = 50, offset = 0).

Starting from v2.55, SelectRangeEvery will also process audio. To keep the original audio, use audio = false.

Examples:

# Selects the frames 0 to 13, 280 to 293, 560 to 573, etc.
SelectRangeEvery(clip, 280, 14)

# Selects the frames 2 to 15, 282 to 295, 562 to 575, etc.
SelectRangeEvery(clip, 280, 14, 2)

this i what im thinking, will let avi:Net do the avisynth and then copy it from the Temp folder and add the SelectEveryRange "line"

kind of select 10 seconds every 5 minutes to encode, then feed that "modified" avisynth script on avi.net by putting this job in first place on the job list.

Thanks

[iNFO-DVD]

Could you have it remember the options across sessions? It does not remember quantizer settings now.I was going to say..... it does remember the options, but that's de-interlace, autocrop e.t.c. I didn't allow it to remember size, bitrate e.t.c. as this will usually change everytime a file is loaded in but I can see that it would be sensible for the Quant to stay at the set value. I will do that now so it will be ready for next version.

been testing 1 pass quality modes/need for comptest since this thread "heated up".Yes I know the comptest does seem to be the 'in' conversation at the moment. I must admit, I am tempted to have a mess. Not sure what I can do with the result from the comptest as my hands are tied a bit because of standalone player support, which I will NOT ditch. But maybe I could, at least for now, do a real quick comptest before encoding to at least give the % value in the display to give the user an idea. That would be better than nothing wouldn't it?

can anyone tell me how i can make a comptestWell I'm going to have to do some serious reading later when I get some time. I done a lot of searching last night and didn't really get anywhere. I have read some really good posts on comptests but just couldn't find them when I needed them, typical. I'm under the impression it's something like:

1. use the SelectRangeEvery option to get 5% of the movie/clip in batches of 14 frames.
2. do a first pass encode. (hmmm, or a single pass q2 encode?????)
3. read information from log file (.pass file)
4. discard first and last of every 14, more with B-frames
5. find the amount of bits, re-calculate back to all of movie.
6. calculate it's equivalent QF which gets divided by original QF to give a value.

This is just of the top of my head, as I now want to implement something into avi.NET I need to do some searching/reading myself to actually find the proper, exact method.

[weaver4]

Please save both the quantizer value and the "Single Pass Quan" checkbox. Thanks.

[iNFO-DVD]

Please save both the quantizer value and the "Single Pass Quan" checkbox. Thanks.Will do :D

[unskinnyboy]

The crux of the test is not just knowing how to use SelectRangeEvery(), but the actual computation of the percentage value.

Reading materials:
http://forum.doom9.org/showthread.php?s=&threadid=44414
http://forum.doom9.org/showthread.php?s=&threadid=29357

And just for an idea as to how other tools use SelectRangeEvery(), GordianKnot does a SelectRangeEvery(280,14) by default and AutoGK does a SelectRangeEvery(300,15). Regardless of what values you use, a minimum of 5% should be sampled to get accurate results. The more the better, but beyond a certain point the test starts hurting the total time taken for the encode as a whole.

[iNFO-DVD]

Can some of you experts help me out here? I just want to get some clarification on my first step, I'm reading conflicting reports.....

(COMPRESSION TEST - FIRST PART)

I'm using 5% of movie/clip and doing 14 frames a shot.
All settings are as if I'm about to do normal convert.
I do a first pass.
I look at, and parse, the video.pass file, which now looks like this:

i 2 680 0 0 1936 1936
b 4 0 680 0 8 8
p 2 0 0 680 98 98
b 4 0 680 0 378 322
p 2 194 79 407 1306 830
b 4 0 680 0 153 144
p 2 224 58 398 2721 984
b 4 0 680 0 538 395
p 2 215 81 384 3771 973
b 4 0 680 0 908 517
p 2 210 100 370 4497 1008
b 4 0 680 0 209 168
p 2 206 120 354 5064 1028
p 2 19 310 351 4168 789

I know what the values mean in each section (good ol' search) and I know I have to discard the first and last line. But as avi.NET is uses 1 b-frame and only ever 1 b-frame (they don't get turned off) I've read different reports of what else needs to be dropped, do I drop more lines? which ones?

TIA

[Sharktooth]

@all: please redirect your questions/replies to this thread: http://forum.doom9.org/showthread.php?t=115381

[iNFO-DVD]

@all: please redirect your questions/replies to this thread: http://forum.doom9.org/showthread.php?t=115381Thanks.