relations between x264 options

[yaous]

1) --cplxblur and --qblur are adjustable in CRF mode.

2) --qpmax/min/step, --ip/pbratio and --chroma-qp-offset are adjustable in CQP mode.

3) AQ is not used in CQP mode.

are these correct?

[nurbs]

3) is correct. Also pbratio isn't used when mb-tree is on.

[LoRd_MuldeR]

1) --cplxblur and --qblur are adjustable in CRF mode.

I think those are 2-Pass only options. Even if they are adjustable, they mean nothing for CRF mode.

Please correct me, if I'm wrong...

2) --ip/pbratio and --chroma-qp-offset are adjustable in CQP mode.

I think so, yes.

3) AQ is not used in CQP mode.

Yes. CQP mode uses constant quantizers, hence AQ doesn't apply here.

Yet another reason to not use CQP ;)

[yaous]

I think those are 2-Pass only options. Even if they are adjustable, they mean nothing for CRF mode.

Please correct me, if I'm wrong...

AFAIK, both crf and 2-pass calculate complexity from previous encode(2-pass) and from SATD residual(crf).

then blur complexity and allocate bits acording to the blured complexity.

is that correct?

if so, --cplxblur may affect crf mode, too..???

I'm sorry for my bad english..

[LoRd_MuldeR]

AFAIK, both crf and 2-pass calculate complexity from previous encode(2-pass) and from SATD residual(crf).

There is no "previous encode" in case of CRF mode.

then blur complexity and allocate bits acording to the blured complexity.

is that correct?

From what I see, it seems the one and only place where 'f_complexity_blur' is used is init_pass2().

[Dark Shikari]

There is no "previous encode" in case of CRF mode.Sure there is. Previous frames.

[LoRd_MuldeR]

Sure there is. Previous frames.

Yes, okay. But there is no knowledge about the frames that have not been encoded yet in the current pass, because there is no data from a "previous encode" for those frames.

[yaous]

From what I see, it seems the one and only place where 'f_complexity_blur' is used is init_pass2().

Thanks for the reply :)

and...in my post

'previous encode' correspond to 2-pass,

'SATD residual' correspond to crf.

sorry for my poor english again.

[RunningSkittle]

Yes, okay. But there is no knowledge about the frames that have not been encoded yet in the current pass, because there is no data from a "previous encode" for those frames.

isnt this what lookahead is for?

[Dark Shikari]

isnt this what lookahead is for?Yes, with lookahead the "complexity" measure between both passes is identical (when MB-tree is on). cplxblur has no effect when MB-tree is on.