At what point is a second pass pointless?

[Arlo]

I'm using AutoGK; at what single pass Target Quality setting (e.g. 75%) does the quality negate the benefits of using two passes for the same filesize?

[manono]

Apples and oranges. Use Target Quality if you don't need to aim for a specific size. Run the 2-pass when you need the AVI to be a certain size. A 1-pass that winds up a certain size will have about the same quality as a 2-pass done for the same size, all other things (like the resolution) being equal, no matter the percentage used. Some might say (and I'm one of them) that the 1-pass will have better quality.

Your question implies that running 2-passes, by definition, produces better quality output. It doesn't.

[Arlo]

Apples and oranges. Use Target Quality if you don't need to aim for a specific size. Run the 2-pass when you need the AVI to be a certain size. A 1-pass that winds up a certain size will have about the same quality as a 2-pass done for the same size, all other things (like the resolution) being equal, no matter the percentage used. Some might say (and I'm one of them) that the 1-pass will have better quality.

Your question implies that running 2-passes, by definition, produces better quality output. It doesn't.

Presumably my assumption that the single pass is produced at a constant bit rate is wrong then?

[manono]

Presumably my assumption that the single pass is produced at a constant bit rate is wrong then?

That's correct, it's wrong. The one-pass mode is a constant quality encode with the quality determined by the percentage chosen. It's very much variable bit rate.

[Arlo]

Presumably my assumption that the single pass is produced at a constant bit rate is wrong then?

That's correct, it's wrong. The one-pass mode is a constant quality encode with the quality determined by the percentage chosen. It's very much variable bit rate.

Cheers, I've just found the doom9 autogk forum whichpointed this out

[Arlo]

manano,

I don't suppose you know how 1 and 2 pass encodes work with Nero Recode 2 (Digital AVC)? Unlike AutoGK, with Nero Recode you specify target filesize for both 1 and 2 passes; I've tried them both on the same input file and 2 passes gets you closer to the target size and one pass is slightly smaller (so still fits on the CD for e.g.). So is it still a case of the quality being the same (for the same filesize) but that for 1 pass it is less accurate as far as the final filesize is concerned?

The reason I ask is that I am not too bothered about filesize for movies and so prefer the Target Quality control of AutoGK (plus I can burn in subtitles) but for TV shows I want to trade some quality for disk space and so want the quality benefit of H264 (as they are more compressed than the movies) but I'm not that bothered about exact file sizes so it isn't worth the time it takes for 2 passes. Hopefully some of that made sense!?

[manono]

So is it still a case of the quality being the same (for the same filesize) but that for 1 pass it is less accurate as far as the final filesize is concerned?

I've never used Nero Recode but if there's a 1-pass for a specific size, while it may (or may not, dunno) be 1-pass VBR, it can't produce the quality of a full 2-pass encode.

If you could see a graph of the bitrate, in a 2-pass encode it fluctuates a lot, depending on the complexity of the various scenes. The bitrate graph of a 1-pass VBR encode just wobbles around the given bitrate and for a CBR encode it's more-or-less a straight line. With both forms static scenes get too many bits and complex scenes are bitrate starved.

A full 2-pass encode uses the first pass to gather the data needed to distribute the available bits optimally, while the 2nd pass writes out the video. If you need the video for a specific size and are at all concerned for quality, go 2-pass.