[demograph68]

Here's a clip I tested at 120 kBytes dual pass for all encoders.

Source MPEG2


H.264


Xvid


3ivx


Sorenson3


Apple-MPEG4

[Uncle Skeleton]

You can't use still frames to compare video codecs. They might not all put keyframes in the same place, and the quality of the post-filtering won't be conveyed in still frames. Also, though less of an issue, you shouldn't use a compressed source (mpeg-2) since you'll be confounding your results with how the test codecs deal with compression artifacts (visible or not) in the source footage.

[bmedina]

And how'd you get dual pass with Apple's encoder(s)? Which H.264 encoder? What was the filesize for each? What were your impressions of the video in motion?

[demograph68]

Apple MPEG4 is the only codec which doesn't offer 2-pass. The 3ivx example can be used without the codec. (simple profile, the ful version is nearly on par with Xvid) I'm using sorenson's AVC/H.264 codec. It would be nice to show a clip of them instead of a frame though. AVC encoding was slooooow while 3ivx was fast. As for viewing it in it's entirety, I'd first recommend H.264, and then Xvid, though 3ivx is almost just as good.

The clip was 2 minutes. The average filesize for them is 14.3 MB. No audio.
I may try try this again but with a better source. (Uncompressed)

[bmedina]

One more question: what did you use to encode to Xvid?

[demograph68]

Originally posted by bmedina:
One more question: what did you use to encode to Xvid?

FFmpegX