View Full Version : What exactly does a 2 pass render do?


Stephen Kettyles
August 31st, 2008, 01:14 PM
Hi folks,

I was just wondering what the two pass render does and does it actually make the quality noticibly different. I always render with two pass but is obviously takes twice as long. Should I bother?

Giroud Francois
August 31st, 2008, 02:13 PM
the first pass is encoding for nothing , just to see how the bandwith react.
a log file is created.
the second pass is encoding for real with the help of the log file, so the encoding knows in advance the bandwith to allow to each frame for best quality/size ratio.
multipass encoding gives not better encoding, just better use of of bandwith and space.
if properly used , you can get better result.
i think that it could be possible to write a multipass encoding into a one pass encoder, just by using a buffer that can calculate few gops ahead and forward the info to the currently encoded gop. I never see such feature while it could be easy to write.

Dan Bridges
August 31st, 2008, 09:20 PM
The first pass determines where the hard, normal & easy parts are situated.

The second pass then encodes, allocating the bit-budget intelligently.

The requested average bitrate should be closely approached using this method and the best balancing of quality versus filesize achieved.

Just looking a few hundred GOPs ahead won't work as well because the encoder won't know how tough/easy the encoding situation will be much further ahead, say another 10 minutes into the clip.

Paul Cascio
September 2nd, 2008, 01:19 PM
Interesting topic. Would a single-pass encoding be biased toward quality (high bit rate) as opposed to file size?

Don Donatello
September 2nd, 2008, 11:33 PM
" Would a single-pass encoding be biased toward quality (high bit rate)"

depends on how you set the bit rate range ... i believe Vegas defaults the mpeg bit rate to Max 8,000,000bps , average 6,000,000bps and Min 192,000bps ... so if a scene is pretty static and not complicated it will go to the lowest bit rate and may miss few to several frames if something complicated /fast motion enter frame quickly ???
i believe the most effecient would be using 2 pass ....
also you can change the max, average , min to whatever you like ( though i found 2,200,000bps constant to be just acceptable for taking heads = low quality daily's = 4 1/2 - 5 hrs on single layer DVD)