Multipass VBR - Worth it?

[quantum]

I've used CCE for hundreds of dvd backups. I finally got around to testing the difference between a 4 pass vbr encode and a 1 pass vbr encode. I did a low bitrate encode at 1800 average figuring this might highlight the differences.

I captured screens at several points throughout the movie and compared them carefully. While some pixels change in the 4 pass, it's very minor. No way would anyone be able to tell on a TV. I was hoping the compression artifacts might be reduced in the 4 pass, but that was not the case.

Has anyone found a real world case where a 3 or 4 pass makes any noticable difference? I'm thinking I'll stick to 1 passes from now on.

[r6d2]

Originally posted by quantum
Has anyone found a real world case where a 3 or 4 pass makes any noticable difference? I'm thinking I'll stick to 1 passes from now on.
As you may know, in theory, multipass optimizes the distribution of BR constraints by rearranging and patching here and there. It is a very time-consuming process, designed mostly to encode when there are no space limitations and you just want the best quality you can get.

But, as they say, "in theory, the difference between theory and practice is much smaller than what it is in practice."

As well as you, in most tests I've had a hard time trying to tell the difference. I guess it requires a trained eye and a HDTV :)

I still, though, use multipass 2+VAF (no longer 3 or 4) for when I just "smell" the movie is too complicated for a single pass.

Lately I've been using Tylo's plugin (search for D2Sroba) for OPV with excellent results. So good that almost unbelievable. For instance, the movie "The Ring", fitted on 1 CD with Q=25, and it is 1:47 long!

(I must confess I cheated using DDogg's suggestion of UnDot() and Deen().)

That movie is very, very compressible, and by no means you could have told that without a measure instrument like D2Sroba.

Hope this helps.