Megui - Undersized XVID file

[skaleton]

HI,
I have a box-set of 8 dvd's containing a 90 mn movie of the same series.
I want to back them up to a 1/4 dvd Xvid avi file
I used Megui as, from the test I made, it needs srestore to convert it from 29.97 to 25 fps
I used the same settings to all of them ( I checked each one before)
7 gave me the 1/4 dvd that I was expecting but one only produced a 800 MB file

I re-ripped it, but with the same result. And I made other conversions in-between and after-wards without any problem

Would someone please help me identify the problem and find a solution?

EDIT: I realise the this post should probably be in the DVD Ripping forum

[manono]

It could very well be that you saturated the codec (got the very best quality and more bits can't be squeezed out, given your settings). You got 800 MB and wanted 1100 MB or so, is that right? Doesn't it look OK to you? I have no idea of the kind of source (animation, live action), how compressible it is (dark and static, bright with lots of motion), the resolution used or anything else, so one can only guess. But if it was set up correctly (like the others), and the others achieved the size you wanted, my money goes on codec saturation. If you really want to increase the size for some reason, use a higher resolution, better quality audio (AC3 instead of MP3, for example), no B-Frames, a better quantisation matrix, a sharpening filter, allow for the use of Quant 1 (a waste of time, if you ask me), etc.

[skaleton]

Thanks manono.
I read about that saturation although it was for the Divx
But what I find weird is that it would happen in a set of live action dvd's that all have the same kind of atmosphere and tempo, kind of soft lighting and 'normal' motion, have the producer, the same distributor, the same average bitrate
and yet one of them, with the same AVS scrypt and the same xvid setting would produce a much smaller size file.
And I expected that saturation would occur at much higher bitrates, not around 1100 kbps if I remember correctly, especially as I use the highest resolution.
What if I set an objective of 1/3 dvd? I suppose that if the codec was saturated it should still top at 800 MB.
What do you think?

[manono]

Yes, without making any other changes, just setting a higher bitrate should also produce the same size if the codec is saturated. You might run them all through DRF analyzer and check the results:

http://www.divx-digest.com/software/drfanalyzer.html

The 800 MB one should have a lower number, where if it could achieve the desired size but didn't for some reason, it should have a higher number than the rest, given that you say all the source material is similar and all the settings are pretty much identical.

[skaleton]

I did a new encode during the night with a 1/3 dvd target and indeed the output is the same as before, 830 MB with a video bitrate of 1086 kbps and a quality factor of 0.151 bpf instead of 1120 MB, 1550 kbps and 0.215 bpf like the other files.

I ran the analysis through DRF Analyser
The average DRF is 2.98 whereas the other files range from 3.2 to 3.4
I attach 2 reports one for the file I am concerned about and the one with the highest result
Would you please advise how to interpret them and how they can help me improve my encodes?

[manono]

One way to improve them (and also to boost the size a little bit), is to adjust your Max Consecutive BVOPs/Quantizer ratio/Quantizer offset numbers. With that 830 MB one, your I and P-frames are all quant 2 (the sure sign of saturating the codec, assuming you don't allow for quant 1), but your B-Frames are all 4. Better would be to have Quant 3 B-Frames when the I and P-Frames are 2. That's just my opinion and I have no idea whether or not others will agree. I have my three numbers at 1/1.50/1.00. I guess yours are at 1/2.00/1.00, or something like that. I don't know if yours are default or not. If I remember, the default is 2 max consecutive b-Frames but don't know for sure now, and I don't feel like setting them back to default just to check.

Apparently that video is much more compressible than the others in the series. Because this app is old, and there are some screwy things about it to begin with, you can ignore the stuff about the recommended resolution. Based on the size of the encodes, the resolution seems fine. The B-Frames increase the average quant which almost always results in it recommending a lower resolution. It's still a useful app for XviD and DivX encodes, I think. Oh, and DRF is Detail Removal Factor, which is another name for Quant or Quantizer.

[skaleton]

Thank you manono
Megui is different from the standard xvid settings so I send them to you.

A far as I can see, it seems that I have the same settings, regarding the B-Frames. Right?

I read, maybe even from you, that the Quantizer range should be reduced to 2-4 instead of 2-31. Should I change it that way?

Although I need a xvid, I did a x264 encode and got a 1120 MB file. So I guess a file may saturate a codec but not another one?
I installed avi.net (seems to be a good program) and see if I get the same kind of xvid file.

[manono]

Yes, many of those settings are similar. I do mostly 1-pass constant quant encodes these days because I usually don't care what the final size is. For 2-pass encodes, yes, I severely restrict the I and P-Frame settings, usually to 2-6, but sometimes to 2-4. I don't think mine has a min/max B-Frame setting.However, you had better know what you're doing if you restrict them that much because you're liable sometimes to get oversized encodes because XviD won't have the ability to use higher quants to keep the size down. Not with this current project, though.

Although I need a xvid, I did a x264 encode and got a 1120 MB file. So I guess a file may saturate a codec but not another one?

I don't encode using x264 so I don't know, but that sounds reasonable.

[skaleton]

Thanks again, manono, for all your help