DivX 5.05 vs 5.1 Compression Test

[BluDChyLD]

Just ran a few compression tests to see how 5.1 competes against the old 5.05. Tests use neutral bicubic resize and a res of 608x336.

5% comp.test with 5.05 = 66.9%

5% comp.test with 5.1 + psy.visuals off = 57%

5% comp.test with 5.1 + fast psy.visuals = 48.1%

5% comp.test with 5.1 + slow psy.visuals = 41.2%

All tests made using bframes, gmc and standard performance settings. I must admit, I am really disappointed by this :( These tests also confirm the larger quality based files using psy. visuals here (http://forum.doom9.org/showthread.php?s=&threadid=60718) .

I think i'll be sticking to 5.05 for now: Even if 5.1 does warrant better quality, overall lower compression will mean encoding at lower resolutions. Plus with my system specs slow and slowest performance settings are out of the question running at 2-5 FPS!

I'm not knocking 5.1, but it seems like they're squeezing more detail and quality out of the codec at the expense of speed and filesize. Im into making 1 cd rips, so this ain't good for me :( Feel free to most your own results...
Hope this helps

[LordIntruder]

Do not stay with an older version only because your comp test displays weirds things. Maybe the comp test option need to be update to comply 5.1? I don't know but one thing I'm sure is 5.1 do not need to encode at slower resolutions to compete with 5.05.

I didn't make any test with psy off and on to see if the sizes were different but I encoded some movies with beta divx and the final target size was kept. I targeted 697Mb I got 697Mb and with Psy slow. So there is no filesize cost to get a better quality. To me 5.1 is 5.05 with a much better overall quality, less macroblocks, etc... And even in standard mode you get a better result than 5.05. Like you i'm making one cd rips. My latest one was an 4:3 PAL movie, 1h39min with 2 audio streams (96K-48Khz MP3 each). 802K final bitrate. The final size was 700Mb, no macroblock at all, a very good quality.

[Sagittaire]

DivX 5.1 is better than DivX 5.05 in standard mode and very better in slowest mode ... Compressibility test isn't quality test ... ;-)

[DigitAl56K]

FYI There are a few comments on a mirror thread here:
http://forums.divx.com/viewtopic.php?topic=53460&forum=6

[LordIntruder]

Today I make some tests with and without PV. I didn't noticed at first that IF you make multipass encodings, PV or not PV the filesize will still be the same or very very close. Those that seems to get increased filesize encode at constant quantizer.

Thanks for the link DigitAl56K, sure we learn every day at every corner of the divx road. :)

[qwprince]

i did see a difference .. only its the other way around .. 5.05 giving better quality at slowest(with less time taken) compared to standard 5.1 (40mins - 1hr more)

i did encode LOTR-Fellowship(179mins) .. and details are much low in 5.1 standard