CQ-ASP_Q2_eq(crf) MeGUI Custom x264/AVC video profiles

[rexrex]

hi,

I got a question about the profiles from Sharktooth.
What is the right filesize for a video? (this is not my question here), but this is a question I always have, when I make a video. The solution is the use of CQ-ASP_Q2_eq(crf) to have a stable quality. But I read once, that the same filesize with 2pass would look better...

So my first Question, is this true?

And if this is true, can someone make a 2pass, where in the first step the filesize is given and after the second pass, I got the best quality, I can achive with this filesize?

TIA

PS: interessting articel about MPEG4-AVC on Blueray:
http://www.anandtech.com/video/showdoc.aspx?i=2886
maybe we can get a new HD profile...

[quake74]

Since you are new, you should have read the rules of this forum recently, in particular the "Do not ask what's best" rule.

The point is that there is no "right filesize" for a video. Most of the times, the bigger the better the video, and the best is just the source! So compressing is always a tradeoff between filesize/quality/timeofcompression. The Q2_eq (internally it means CRF 18) profile gives you an almost indistinguishible version of the original video to me, but in my backups I use CRF 24 because it's much smaller and good enough for my eyes. Using these CRF modes, you have no control on the final filesize of the video. If you need to put that video on a CD or DVD, then you want to use a 2 pass mode, which makes certain that the file will fit, at the expense of a much longer time of compression. Actually, you could do 3, or 4, or 10 passes, and each time it will be a little bit better, but for me it's a waste of my cpu time.

[rexrex]

@quake74 you got me wrong,

I don`t want the "perfect filesize/bitrate" ;) , again I want to know, if the video with the filesize that the profile CQ-ASP_Q2_eq(crf) "made" would look better in twopass mode, then in the default onepass of the profile. I don't care of a lil bit, should be significant...I'am not new to videoencoding, but new in this forum. btw I was to lacy to find out by myself, maybe some here tested this already and know the answer. If I get you right, you mean, that it will make no big difference, so I dont need 2 pass... TIA

[quake74]

again I want to know, if the video with the filesize that the profile CQ-ASP_Q2_eq(crf) "made" would look better in twopass mode, then in the default onepass of the profile. I don't care of a lil bit, should be significant...TIA

Yes, the two pass mode using the same bitrate that the Q2 ends up using will look better. If you make a 3 pass, it will look even better and so on for 4,5,6... From 1 to 2 it will be easy to see if you know where to look (specific high motion scenes, low contrast scenes). From 2 upward it will be really hard.

[rexrex]

...is it significant? you know for sure?

[check]

there is an improvement of 2pass over CRF/CQ. See this post for a qualitative estimation of the increase: http://forum.doom9.org/showthread.php?t=116773&page=3#50

more than two passes is probably overkill though - the increase in time compared to the increase in quality is very large :)

[rexrex]

ok thanks but now I have do the films 2 times, first crf to know the filesize and second one 2 pass for final encoding...

[quake74]

Rather than doing that (which is a waste of a lot of time) read the whole thread that check pointed you to. Or compress just 10% of the movie with constant crf and then multiply for 10 (you can use a statement like SelectEveryRange(2000,20), look for it in avisynth forum). Good luck and good encoding!

[rexrex]

thanks quake74, finaly... but I know, if I make just 10%, other might say, you know nothing with 10% :) I had this discussion allready and I am not really in avisynth, thats why I am using megui or stax ( I wish they could easily live togehter :) )