Which gives smoother video constant quality or 2 pass?

[thebach]

which delivers smoother video in TMPG, constant quality or 2 pass? I don't care about file size and if constant quality gives video as smooth as 2 pass I'm gonna use it since the encoding time is half. Damn I meant to put this in the advanced conversion forum

[adam]

If by smooth you mean the movement than it really shouldn't matter, the encoding modes effect how the bitrate is distributed which will just effect overall quality. Both modes work well, and different people have different preferences. Logically, 2-pass should always provide better quality, and in my opinion it does. However, in many respects, TMPGenc is not the most logical encoder and ultimately the better algorithm is the one that is better programmed. 2-pass is much more consistent and much easier to use, but it takes twice as long. You can probably get comparable results in a much shorter amount of time with CQ, but if you overshoot the filesize than you will have start all over, and you have negated the speed increase, and if you undershoot the filesize than you have lost some potential quality.

DVD2SVCD has a built in prediction test to allow you to better set your bitrate and there is a similar script at www.kvcd.net which you can use to manually run such a test. This takes alot of the guesswork out of using CQ, but you still run the risk of over or undershooting the desired filesize.

I'm moving this post to advanced conversion since that's where you wanted it.

[kwag]

I
there is a similar script at www.kvcd.net which you can use to manually run such a test. This takes alot of the guesswork out of using CQ, but you still run the risk of over or undershooting the desired filesize.

Adam,

The currect KVCD file size prediction formula, when used with KVCD's Q. Matrix and GOP, hits ~-2% from wanted target almost every time. It never goes over. It's as accurate as (or more than) CCE's x-pass or TMPEG's 2-pass.

-kwag

[adam]

%2 is great, almost negligable. If this is the case you may very well have exceeded the file predictability of TMPGenc's 2 pass VBR, since in my opinion its output is very inconsistent. CCE however is %100 percent accurate, since it maintains the exact avg you set you get no variation. I can literally hit exactly 800MBs on the head everytime. Keep up the good work though, accurate file prediction for 1-pass VBR is certainly a welcome thing.