Some quick x264 questions

[Taktaal]

I was trying to understand a bit more about x264 options than just pressing a button in MeGUI and happened across some questions. Some could be answered by myself by looking through the forums, I'll include those here for the benefit of other people reading this thread :p
Maybe someone else can answer the others.


What happened to the --bime option from earlier versions that is used so much in sample configs?
It was replaced by the --subme option. If that value is set to 1-4, bime is off, if subme is 5-9,then bime is automatically on
What happened to --fpel-cmp?
Replaced by "--me tesa". It's still really slow and apparently doesn't give much quality


And the questions I didn't find an answer for

Most older sample configs like from http://forum.doom9.org/showthread.php?t=130786 use --bframe 16 even on low quality encodes. But recently I've read that --b-adapt 2 is a pretty good setting to have. Unfortunately the two are really incompatible because of speed hits. Should we encode with lower numbers of bframes nowadays? Does b-adapt actually do anything on a single pass encode?
Is there any benefit of using two pass encoding when you're not that interested in file size? Simply said, if I encode a file with "-q x" and it happens to be 500MB big, then do it again with a two-pass encode and target the result to be exactly of the same size, will it be better quality due to somehow getting more analysis?
What's the deal with the cqm matrices. I understand that they somehow provide weighting to some algorithm as to what decisions to pick, but how do I optimize this? Are there like ideal matrices for slow movies, action movies, cartoons etc
Are p4x4 partitions ever worth it? In my tests, they really slow the encodes down and don't seem to add any quality at all...

[Dark Shikari]

Most older sample configs like from http://forum.doom9.org/showthread.php?t=130786 use --bframe 16 even on low quality encodes. But recently I've read that --b-adapt 2 is a pretty good setting to have. Unfortunately the two are really incompatible because of speed hits. Should we encode with lower numbers of bframes nowadays? Does b-adapt actually do anything on a single pass encode?Of course b-adapt "does something" on a single pass encode; otherwise what would be the point?

Yes, b-adapt 2 with fewer b-frames (2-3) is better than b-adapt 1 with more (16).
Is there any benefit of using two pass encoding when you're not that interested in file size? Simply said, if I encode a file with "-q x" and it happens to be 500MB big, then do it again with a two-pass encode and target the result to be exactly of the same size, will it be better quality due to somehow getting more analysis?You should never use constant quantizer mode, ever. But there is no point to 2pass if you don't care about a specific size, no. That's what CRF is for.What's the deal with the cqm matrices. I understand that they somehow provide weighting to some algorithm as to what decisions to pick, but how do I optimize this? Are there like ideal matrices for slow movies, action movies, cartoons etcThe "ideal matrix" in most cases is none at all. If you don't know what it is, don't mess with it.
Are p4x4 partitions ever worth it? In my tests, they really slow the encodes down and don't seem to add any quality at all...They're helpful at higher bitrates and/or lower resolutions.

[Esurnir]

They're helpful at higher bitrates and/or lower resolutions.

And lower CRF ?

[Sharktooth]

higher bitrate = lower average quantizer... so even lower CRF...

[Taktaal]

thanks a lot for the quick answers and happy new year everyone :D