R-Lambda and CQP comparison

[xkfz007]

Why is the R-Lambda RC method worse than the Constant QP method in the HM11.0rc1 in my test?
Rcecently I have done some tests about the R-Lambda RC method. But compared with the CQP method, the BD-RATE of R-lambda is worse.

[benwaggoner]

Could you give some more context here? Are you just saying that the bitrate is lower, or is there also a perceptual quality difference.

Also, your build is pretty old. There were final 11.0 and 11.1 releases, and HM 12.1 is the current version. There may have been salient changes since your build.

[xkfz007]

Originally Posted by benwaggoner

Could you give some more context here? Are you just saying that the bitrate is lower, or is there also a perceptual quality difference.

I calculated the BD-RATE between the R-lambda and CQP for about 14 sequences @720p and 1080p. The final BD-RATE was about 9.81%.
And then I searched some papers about Rate control and I learned that the CQP method encoder was better than encoder with the RC method.

Originally Posted by benwaggoner

Also, your build is pretty old. There were final 11.0 and 11.1 releases, and HM 12.1 is the current version. There may have been salient changes since your build.

I have read the HM12.1 source codes, in which there is only the R-lambda RC method left and the other one in HM11.0 is deleted. And the R-lambda source codes are almost the same.