Log in

View Full Version : Question about x264 1-pass rate control arithmetic


clu
30th October 2007, 03:28
I‘m studying x264 rate control arithmetic.In 1-pass,i can't get clearly about that how to use SATD value to get qp.There is any document can tell me the process about it?

i try to use a pointing DSP to achieve the 1-pass arithmetic,so i want to know is there a pointing arithmetic can take place of the floating arithmetic original?

TKS and Sorry for my poor English.

xiejun06
30th October 2007, 12:35
I am also study the rate control part of x264. After reading a article of Morreti(if you need, I can provide to you), I think this part maybe different from JM. According to the article, x264 use 5 methods of rc, which is from ffmpeg. I also want some detail information, maybe we can discuss about rc.

clu
31st October 2007, 02:05
i have got the “A qualitative overview of x264's ratecontrol methods By Loren Merritt”,and i have migrated the 1-pass airthmetic to DSP, but i dont know why we can predict qp used satd? is there any paper for reference

xiejun06
2nd November 2007, 08:52
About the MAD and SATD, I remeber a paper said when using RDO, the performance of SATD is little better. The article you mentioned, I have read also. It don't have any detail about the alogrithm, I want know why set the parameter like this, why use this method, and so on.

clu
21st November 2007, 09:47
Maybe we got the same problem and both us wanted to know why x264 used this arith. Is it just based on exp?

xiejun06
3rd December 2007, 16:13
I am also working on the rate control on DSP, and I have some problem about migration, would you provide me some help? thanks.
my email: xj19830801@163.com

Dark Shikari
3rd December 2007, 16:27
x264's ratecontrol is much like a frame-based adaptive quantization; more complex frames (as judged by SATD) are given higher QPs, while less complex frames are given lower QPs.

Sagekilla
4th December 2007, 00:14
x264's ratecontrol is much like a frame-based adaptive quantization; more complex frames (as judged by SATD) are given higher QPs, while less complex frames are given lower QPs.

Does this apply likewise to constant rate factor mode, in that x264 will allocate less bits towards a complex scene and more for the relatively calm ones?

Dark Shikari
4th December 2007, 01:13
Does this apply likewise to constant rate factor mode, in that x264 will allocate less bits towards a complex scene and more for the relatively calm ones?Yes. The only mode that doesn't use this approach is CQP.