XVID conversions too small

[jagabo]

Oh sorry, I forgot: your video is that odd one that looked like normal Divx/MP3 AVI but there was something weird about file. Yes, just reencoding with AviDemux is probably your easiest solution.

The source file's audio is only 128 kbps. Reencoding at 320 kbps is overkill. But it won't hurt. The default Xvid settings are reasonable. On the Frame tab you might change the Max b-frames to 1 and select Closed GOP and Packed Bitstream for better Divx compatibility. On the Quantizer tab the MPEG Matrix will be slightly sharper than h.263. But it will also raise the bitrate a bit.

[coyote2]

Oh sorry, I forgot: your video is that odd one that looked like normal Divx/MP3 AVI but there was something weird about file. Yes, just reencoding with AviDemux is probably your easiest solution.

I unfortunately complicated, and may have confused, matters, by needing help with two different files this thread. The first one (work is now completed on) was an .avi file (DIVX/mp3) about which jagabo noted "there's something odd about the way it's encoded".

The file I'm now seeking help with is a .m4v file I first mentioned here.

Does this possible clarification change your advice?

For background, for dealing with the .m4v file jagabo suggested

• an AviSynth/VirtualDub approach (which I haven't tried yet even though jagabo posted some script code, because AviSynth doesn't seem easy to this noob)
• Haali (which created a MKV file I typed ".avi" as an extension for), and most recently
• AviDemux (today's remuxes wouldn't play on my 5992, but a recode does)

The source file's audio is only 128 kbps. Reencoding at 320 kbps is overkill. But it won't hurt. The default Xvid settings are reasonable. On the Frame tab you might change the Max b-frames to 1 and select Closed GOP and Packed Bitstream for better Divx compatibility. On the Quantizer tab the MPEG Matrix will be slightly sharper than h.263. But it will also raise the bitrate a bit.

Thank you very much. I'm truly humbled by all your help, and wish I knew how to repay everyone. Other than that I promise I will "pay it forward", and will remain dedicated to helping people in the non-video areas I already had some expertise.

[jagabo]

I wonder if changing the Audio Codec ID to 55 would get the audio to play? I don't know how you would do that though.

Since re-muxing isn't working, and re-encoding with AviDemux is working for you, I would just go with that.

[jagabo]

Duplicate post.

[davexnet]

Between 3 - 4 is OK. However, on my Samsung standalone, sometimes it stutters when I use CQ mode.
I can encode it in 2-pass mode, using the same bitrate derived from the CQ mode and it plays normally.
Not sure why, but I think the CQ creates a slightly bigger bitrate spike on high-demand area's.

[coyote2]

Between 3 - 4 is OK. However, on my Samsung standalone, sometimes it stutters when I use CQ mode.
I can encode it in 2-pass mode, using the same bitrate derived from the CQ mode and it plays normally.
Not sure why, but I think the CQ creates a slightly bigger bitrate spike on high-demand area's.

Thank you very much, davexnet! Since my goal is the widest possible compatibility, I think I'll do as you suggest. One question:

When you say "using the same bitrate derived from the CQ mode", do you mean OVERALL bitrate (or just VIDEO bitrate)?
(I ask because I'm uncertain which type of bitrate is meant by the box I'll type it into.

[jagabo]

The 5992 can handle pretty high bitrates. You shouldn't have any problem with target quantizer of 3. Unfortunately, Xvid doesn't have a bitrate peak limiter in target quantizer mode. Divx does though.

[coyote2]

In "Xvid Configuration", is the field "Average Bitrate" for the output file overall, or for only it's video, please?

(I ask because perhaps I should solve my cluelessness about what to enter into this field, with davexnet's mention of matching the bitrate of file output in Quantizer mode).

[jagabo]

In "Xvid Configuration", is the field "Average Bitrate" for the output file overall, or for only it's video, please?

Only the video.

[poisondeathray]

sorry nevermind

[coyote2]

In "Xvid Configuration", is the field "Average Bitrate" for the output file overall, or for only it's video, please?

Only the video.

Thank you very much, jagabo!

Is there a better way to choose an "Average Bitrate", than to match that of the file output in Quantizer mode, please?

(Up thread VirtualDub's Quantizer mode was first mentioned, when I admitted that it's bitrate Calculate button's Filesize options--considering that I have little interest in either conserving, or in simply wasting, space--leave me speechless.)

[manono]

Is there a better way to choose an "Average Bitrate", than to match that of the file output in Quantizer mode, please?

I don't know if it's 'better', but use a bitrate calculator to choose the bitrate for the final size you want. Otherwise, one ordinarily runs a quick compression test to get an idea of the quality of the final size/bitrate chosen. The all-in-one program AutoGK runs a compression test automatically and is quite good at matching up final filesize with quality. It aims for a quality of slightly better than quant 3. You might try it to see if you like the results. It takes IFO, VOB, or AVI input.

[coyote2]

Is there a better way to choose an "Average Bitrate", than to match that of the file output in Quantizer mode, please?

I don't know if it's 'better', but use a bitrate calculator to choose the bitrate for the final size you want.

Thank you very much for your reply, manono!
But since I don't care much what the size of the output file will be (I have little interest in either conserving, or in simply wasting, space), so...

Otherwise, one ordinarily runs a quick compression test to get an idea of the quality of the final size/bitrate chosen. The all-in-one program AutoGK runs a compression test automatically and is quite good at matching up final filesize with quality. It aims for a quality of slightly better than quant 3. You might try it to see if you like the results. It takes IFO, VOB, or AVI input.

This sounds great. One question: is this suggestion to run AutoGK on the input file, or on the output file? If the answer is input, some of the files I'm recoding are .m4v, which I notice isn't a format you mention AutoGK accepts.

[manono]

One question: is this suggestion to run AutoGK on the input file, or on the output file? If the answer is input, some of the files I'm recoding are .m4v, which I notice isn't a format you mention AutoGK accepts.

Right, and that's why I mentioned the formats it accepts. You could run a compress test of your own with, say, 5% of the video, but it'll require knowing some AviSynth. A line such as:

SelectRangeEvery(280,14)

will encode 5% of the whole video, with the snippits taken every 280 frames. Encode for quant 3, multiply the file size result by 20 and subtract a little bit (to account for a large I-Frame being set every 14 frames), to get a rough idea of the final size if encoding for quant 3. If it's too much or too little, raise or lower the resolution, adjust the quality of the audio, filter the video a bit to make it more compressible, change the quantisation matrix used, etc. Finally, choose a bitrate that'll give you the final size you want and run the two 'real' passes.

You could also open an AviSynth script for that M4V video in Gordian Knot and then run the compression test within that program (which also can easily create XviD AVIs).

[coyote2]

The all-in-one program AutoGK runs a compression test automatically and is quite good at matching up final filesize with quality. It aims for a quality of slightly better than quant 3.

Where do I find this "compression test", or it's results, please?

(I installed and ran the program (on one of my .avi files), and I couldn't find either anywhere including the Tutorial, the Advanced Settings, or the Hidden Options. [Unless, if I understood, if one is changing the resolution--which I don't plan to.])

[manono]

Post the log and I'll point it out to you if you can't find it yourself. An entire log will look something like this:

[4/2/2010 8:46:06 PM] AutoGK 2.55
[4/2/2010 8:46:06 PM] OS: WinXP (5.1.2600).2
[4/2/2010 8:46:06 PM] Job started.
[4/2/2010 8:46:06 PM] Input file: E:\VIDEO_TS\VTS_01_0.IFO
[4/2/2010 8:46:06 PM] Output file: E:\Temp\family.avi
[4/2/2010 8:46:06 PM] Output codec: XviD
[4/2/2010 8:46:06 PM] Audio 1: English AC3 2ch
[4/2/2010 8:46:06 PM] Format: AVI
[4/2/2010 8:46:06 PM] Target size: 700Mb
[4/2/2010 8:46:06 PM] Custom resolution settings: fixed width of 640 pixels
[4/2/2010 8:46:06 PM] Audio 1 settings: VBR MP3 with average bitrate: 48Kbps
[4/2/2010 8:46:06 PM] Started encoding.
[4/2/2010 8:46:06 PM] Demuxing and indexing.
[4/2/2010 8:50:45 PM] Processing file: E:\VIDEO_TS\VTS_01_1.VOB
[4/2/2010 8:50:45 PM] Processing file: E:\VIDEO_TS\VTS_01_2.VOB
[4/2/2010 8:50:45 PM] Processing file: E:\VIDEO_TS\VTS_01_3.VOB
[4/2/2010 8:50:45 PM] Source resolution: 720x480
[4/2/2010 8:50:45 PM] Found NTSC source.
[4/2/2010 8:50:45 PM] Source aspect ratio: 16:9
[4/2/2010 8:50:45 PM] Source seems to be pure FILM.
[4/2/2010 8:50:46 PM] Output will contain 157686 frames
[4/2/2010 8:50:46 PM] Decoding audio.
[4/2/2010 9:12:13 PM] Normalizing audio.
[4/2/2010 9:34:33 PM] Encoding audio.
[4/2/2010 9:52:10 PM] Forcing encoding with sharp matrix.
[4/2/2010 9:52:10 PM] Audio1 size: 38,763,562 bytes (36.97 Mb)
[4/2/2010 9:52:15 PM] Overhead: 6,307,456 bytes (6.02 Mb)
[4/2/2010 9:52:15 PM] Video size: 688,932,182 bytes (657.02 Mb)
[4/2/2010 9:52:15 PM] Running compressibility test.
[4/2/2010 10:00:31 PM] Duration was: 8 minutes 16 seconds
[4/2/2010 10:00:34 PM] Speed was: 15.89 fps.
[4/2/2010 10:00:35 PM] Compressibility percentage is: 31.32
[4/2/2010 10:00:35 PM] Using softer resizer.
[4/2/2010 10:00:36 PM] Chosen resolution is: 640x352 ( AR: 1.82 )
[4/2/2010 10:00:36 PM] Predicted comptest value is: 32.49%
[4/2/2010 10:00:36 PM] Running first pass.
[4/2/2010 11:55:32 PM] Duration was: 1 hour, 54 minutes 55 seconds
[4/2/2010 11:55:35 PM] Speed was: 22.87 fps.
[4/2/2010 11:55:59 PM] Expected quality of first pass size: 32.90%
[4/2/2010 11:55:59 PM] Running second pass.
[4/3/2010 2:51:12 AM] Duration was: 2 hours 54 minutes 56 seconds
[4/3/2010 2:51:13 AM] Speed was: 15.02 fps.
[4/3/2010 2:51:13 AM] Job finished. Total time: 6 hours 5 minutes 6 seconds

and I've boldfaced the parts pertaining to the compression test it runs.

[coyote2]

Post the log and I'll point it out to you if you can't find it yourself. An entire log will look something like this:

[4/2/2010 9:52:15 PM] Running compressibility test.
[4/2/2010 10:00:31 PM] Duration was: 8 minutes 16 seconds
[4/2/2010 10:00:34 PM] Speed was: 15.89 fps.
[4/2/2010 10:00:35 PM] Compressibility percentage is: 31.32

and I've boldfaced the parts pertaining to the compression test it runs.

I don't see those lines in my log:

[11/2/2011 7:27:46 PM] AutoGK 2.55
[11/2/2011 7:27:46 PM] OS: WinXP (5.1.2600).2
[11/2/2011 7:27:46 PM] Job started.
[11/2/2011 7:27:46 PM] Input file: C:\work\curtis-NEW\VDub source recode to XviD\25MillionPounds(onlyVLC).avi
[11/2/2011 7:27:46 PM] Input codec: DivX 4
[11/2/2011 7:27:46 PM] Source duration: 52mn 24s 480ms
[11/2/2011 7:27:46 PM] Output file: C:\work\curtis-NEW\VDub source recode to XviD\25MillionPounds(onlyVLC)_agk.avi
[11/2/2011 7:27:46 PM] Output codec: XviD
[11/2/2011 7:27:46 PM] Audio 1: 64 Kbps CBR MP3 2ch
[11/2/2011 7:27:46 PM] Subtitles: none
[11/2/2011 7:27:46 PM] Format: AVI
[11/2/2011 7:27:46 PM] Target quality: 75%
[11/2/2011 7:27:46 PM] Audio 1 settings: Auto
[11/2/2011 7:27:46 PM] Started encoding.
[11/2/2011 7:27:46 PM] Source resolution: 352x288
[11/2/2011 7:27:46 PM] Source fps: 25
[11/2/2011 7:27:46 PM] Analyzing source.
[11/2/2011 7:34:02 PM] Source has percentage of interlacing in motion areas: 0.00
[11/2/2011 7:34:02 PM] Source is considered to be progressive.
[11/2/2011 7:34:02 PM] Output will contain 78612 frames
[11/2/2011 7:34:02 PM] Demuxing audio.
[11/2/2011 7:34:06 PM] Using VAQ in XviD
[11/2/2011 7:34:06 PM] Running single pass encoding.
[11/2/2011 7:34:06 PM] Writing the following script to C:\work\curtis-NEW\VDub source recode to XviD\agk_tmp\25MillionPounds(onlyVLC)_agk_movie.av s
================================================== =========
LoadPlugin("C:\PROGRA~1\AutoGK\DGMPGDec\DGDecode.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\autocrop.dl l")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\RemoveGrain SSE2.dll")

movie = DirectShowSource("C:\work\curtis-NEW\VDub source recode to XviD\25MillionPounds(onlyVLC).avi",25).KillAudio()
movie = isRGB(movie) ? ConvertToYV12(movie) : movie
movie = isYUY2(movie) ? ConvertToYV12(movie) : movie
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples= 10,aspect=0,threshold=0,samplestartframe=0,leftadd =0,rightadd=0,topadd=0,bottomadd=0)
fixed_aspect = 1
c_width = width(cropclip)
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
input_par = (input_par > 1.4) || (input_par < 1.25) ? input_par : (4.0/3.0)
out_width = 352
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples= 10,aspect=new_aspect,threshold=0,samplestartframe= 0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
================================================== =========
[11/2/2011 7:55:14 PM] Duration was: 21 minutes 6 seconds
[11/2/2011 7:55:14 PM] Speed was: 62.05 fps.
[11/2/2011 7:55:14 PM] Job finished. Total time: 27 minutes 28 seconds

I'm hoping you can see what I need to do to change that, please.

[jagabo]

You need to switch to 2-pass VBR encoding if you want to see the compression test.

[coyote2]

You need to switch to 2-pass VBR encoding if you want to see the compression test.

Great, thank you very much jagabo!

One last question; if it says "Compressibility percentage is: 31.32", what specifically does that tell me to proceed to do?

My guess: Run it again with "Target quality" = 69 percent
because 69 = 100 - 31

Edit: I see that ""Target Size" performs 2-passes encoding"; since my not caring much about output size is what led to my quandry about what bitrate to use, I don't see how this approach which manono suggested is fruitful for me.

[jagabo]

I don't use AutoGK so I don't know what the compressibility percentage report is telling you.

The only way to limit bitrate peaks in Xvid is to use bitrate based encoding.

With a target quantizer of 3 you will rarely see peaks the 5992 can't handle, especially from a USB drive where it can handle ~20 Mb/s. I wouldn't worry about it.

[davexnet]

I don't use AutoGK too often, but I think it works best if you leave the resolution to auto,
and set a "custom size" . It allows better compatibility with standalone player, an allows the
"compressibility test" to work properly.

[jagabo]

Here are the opening credits from Big Bang Theory. This is about the toughest video you'll ever run across -- over 16 Mb/s peaks according to Bitrate Viewer (in "per second" mode, even higher in GOP mode). My 5992 has no problems playing this off a USB thumb drive. Use Bitrate Viewer to check the bitrates of some of your videos encoded at full frame size and TQ=3. You'll rarely see peaks over 5 or 6 Mb/s.

[coyote2]

The only way to limit bitrate peaks in Xvid is to use bitrate based encoding.

With a target quantizer of 3 you will rarely see peaks the 5992 can't handle, especially from a USB drive where it can handle ~20 Mb/s. I wouldn't worry about it.

Thank you very much, gentlemen!

True, my 5992 can play the Quant 3 file. But my goal is to create a version everyone can play (I could already watch a different version of the film on my other standalone); so when davexnet mentioned that his Samgung has issues with Quant 3 output, but plays the same bitrate (as he got with Quant 3 output) 2-pass output, I thought I'd do just that: run Quant 3, then plug it's output's bitrate into a 2-pass run to create the final product.

Interestingly (to me), when I ran Quant 3 on my three m4v files (each of which have video bitrate = 1000), the Quant 3 output of the files had quite different video bitrates: 997, 1190, and 959. I was so surprised by that 1190, I ran it again.

[jagabo]

Interestingly (to me), when I ran Quant 3 on my three m4v files (each of which have video bitrate = 1000), the Quant 3 output of the files had quite different bitrates: 997, 1190, and 959. I was so surprised by that 1190, I ran it again.

Of course. Different videos compress differently. There can easily be a 2x or more difference between two videos. See the videos on this post:

https://forum.videohelp.com/threads/295672-A-problem-for-video-experts?p=1811057&viewfu...=1#post1811057

There's a 20 fold difference between the two videos encoded at TQ4.

[davexnet]

Jagabo, Nice looking demo video ("big bang") . What was the source? Over the air ATSC ? Commercial DVD ?

[jagabo]

The Big Bang Theory clip was from a commercial DVD. The MPEG 2 video peaks at around 7500 kbps.

[manono]

I don't see those lines in my log:

Right, I didn't know (or forgot) that you're using an AVI as a source. In addition, as mentioned, there's no point in running a compress test for a single-pass encode. This sort of begs the question, why reencode a downloaded AVI using inferior parameters (limiting bitrate spikes for those few older DVD/MPEG-4 players that will stutter sometimes during complex scenes)? It's already been heavily compressed and you're not doing it any favors by compressing it yet again. It already looks and sounds horrible, I'm sure, and I'm fairly certain at that resolution bitrate spikes of the kind that might affect players won't occur. And doing this for other people, when it plays fine on your player?

By the way, to limit those bitrate spikes and make the results otherwise compatible with the widest variety of players out there as possible), it's best that you turn on the ESS chipset option. You can do that either in the Hidden Options (CTRL/F9) or by reinstalling AutoGK with that option turned on.

One last question; if it says "Compressibility percentage is: 31.32", what specifically does that tell me to proceed to do?

It doesn't tell you anything. It tells the guy whose log I grabbed that the settings AutoGK uses to do it's compress test have to be adjusted and it changes them (resolution, matrix used, resizer) if possible, in order to bring that percentage up to around 70%. But since the guy defined a size and width, that pretty much crippled AutoGK and he was left with a very poor result.

[coyote2]

I don't see those lines in my log:

Right, I didn't know (or forgot) that you're using an AVI as a source. In addition, as mentioned, there's no point in running a compress test for a single-pass encode. This sort of begs the question, why reencode a downloaded AVI using inferior parameters (limiting bitrate spikes for those few older DVD/MPEG-4 players that will stutter sometimes during complex scenes)? It's already been heavily compressed and you're not doing it any favors by compressing it yet again. It already looks and sounds horrible, I'm sure, and I'm fairly certain at that resolution bitrate spikes of the kind that might affect players won't occur. And doing this for other people, when it plays fine on your player?

A different (.mp4) version of it plays fine on my other standalone. The versions I downloaded (from archive.org) are the only versions available, and neither one will play on most standalone players (because many standalone models don't play .mp4's), because "there's something odd about the way [the one that's an .avi is] encoded". My goal is to create a version of this film which will allow everyone viewing on a standalone, instead of on a computer, to be able to watch the film.

And yes, in fact it does already look horrible; but that's true of all available versions. Incidentally, I'm creating a collection of every film by this documentarian; oh, and the .mp4 has a lower bitrate than the .avi I'm recoding.

One last question; if it says "Compressibility percentage is: 31.32", what specifically does that tell me to proceed to do?

It doesn't tell you anything.

I meant to ask, what if my result said that. I guess I should have substituted "would" instead of "does", sorry!

It tells the guy whose log I grabbed that the settings AutoGK uses to do it's compress test have to be adjusted and it changes them (resolution, matrix used, resizer) if possible, in order to bring that percentage up to around 70%. But since the guy defined a size and width, that pretty much crippled AutoGK and he was left with a very poor result.

[coyote2]

Between 3 - 4 is OK. However, on my Samsung standalone, sometimes it stutters when I use CQ mode.
I can encode it in 2-pass mode, using the same bitrate derived from the CQ mode and it plays normally.
Not sure why, but I think the CQ creates a slightly bigger bitrate spike on high-demand area's.

Thank you very much for your reply, davexnet!

I'm curious about the files your Samsung stutters on. It turns out that the files I'm encoding in CQ3 mode, result in video bitrates between 959 and 1190. I wonder if those numbers are low enough, that you think your Samsung wouldn't stutter on them?

(Unless you're pretty confident they'd be stutter-free, I think I'll use those CQ3 output bitrates, to do a 2-pass recode of the source files.)

[davexnet]

959-1190 kbps? That must be average, not peak.
The files I had problems with were all scenes where the small details increased and/or movement resulting in a spike.
There was a scene close to the beginning of Aeon Flux if I remember correctly. The peaks were probably close to 8000 kbps.
My Samsung is CD/DVD only, no USB.
I did post a Demo file, (not sure where I originally got it but I put it up on Mediafire a while ago) where you see a flock
of birds in the sky, and the camera pulls back and the amount of birds increases exponentially. Might be a good test.
It's also a good test to see if your PC can play it with out issues!

Here's the link if you'd like to try it, it's 149MB :
http://www.mediafire.com/download.php?tdukvcaj87nivn7