Maybe it's not worthwhile but I'm going to post some tips and guidelines in response to some problems I've noticed with divx 5.2.1 and previous, and xvid.
When using the quantizer, always use fastest mode setting in divx 5. The quantizer is deliberately borked with the slower modes. This might be true with newer revisions. Even at the maximum value of 1 somehow the quantizer doesn't respond properly to the "motion search" and reduces quality; it's very deceiving since the resulting file is smaller which can fool the user to think the results have better compression when it isn't true. The corrupt results are visually ugly moving blobs in dark sections, and dissipated frequency (smeared and blurry picture) which is typical of very low bitrate. The mode setting of "fastest" disables "motion search" and produces proper expected quantizer results.
Xvid has the same problem, but enabling the trellis quantization option avoids this problem. I'm not sure how Xvid responds without "motion search" or "trellis quantization" but the quantizer might work this way too. Many sources claim trellis quantization lowers quality but this simply isn't true if using quantizer with "motion search".
Some other options in Divx 5 can dissipate the frequency sampled. "Post Processing" should be disabled or only used at the two lowest settings "light" or "standard"; any higher options will lose some frequency. This is mentioned in the manual. Ironically you might want to use higher settings when targeting a small size with a two pass encode, like 90 minutes of qxga on 700 megabytes. If you know the quality will be low, this can help trade off ugly noise for something more pleasing.
It seems the biggest quality problem is the bitrate. If sampling a very noisy source, it's going to need more bits. The only useful way to use divx 5 slower settings, "fast" "standard" "slow" and "slower", and Xvid "motion search" options is with two pass encoding. Divx 5 noticeably smoothes corrupt frequency without losing details with the slower settings so the target size is better accomplished. It's difficult for me to notice an improvement with Xvid. Usually I don't use these options since they are very slow (even a quad core excavator slows down to 7 frames per second).
According to the Divx 5 changelog, vga samples at 30 updates per second are full speed on a 466mhz Pentium 3 with %80 cpu used.
Most bluray players will play divx and xvid content. About 6-7 full length videos 90 to 120 minutes each of standard definition ntsc will fit on a single layer bdr with quantizer at about 3 or 2.8. Higher bitrates are very similar quality to h263.
Divx 5.2.1 seems to work in Windows 95C, but you need at least Directx 8.1 installed.
Divx 3 with ffdshow usually gives the same quality.
There's probably some other stuff I forgot to mention.
Regards
Tips and guidelines for older divx and xvid compression
Re: Tips and guidelines for older divx and xvid compression
Why would you encode video with divx for use on a bluray player? All bluray players support h.264 and VC-1, both of which are worlds better.
I could understand if you were trying to encode videos for some retro playback device, but I'm not seeing the point for playing back on h.264-capable devices.
I could understand if you were trying to encode videos for some retro playback device, but I'm not seeing the point for playing back on h.264-capable devices.