x264: does --no-psnr and --quiet affect quality?

[madoka]

I suspect they do not, but just to want to make sure. Disabling PSNR calculations gives a small but noticable boost in encode speed. At 5 fps, every little bit counts...

[Manao]

They don't affect quality.

[omion]

I'm getting bit-by-bit identical files with --no-psnr (except for the creation times of course).

Strangely enough, on my last test I did, the first pass went slightly slower with --no-psnr. It was around 44.5fps with no PSNR, and ~45.0fps with PSNR on. I assume it's a fluke since the video was so immensely simple. The encoding speed was a bit on the high side anyway...

[omion]

Wait a second... I only get bit-identical results on the first pass. The second pass is not identical. I did a 200-frame test of it, and only about half the frames are identical. A few of the frames even use different quantizers. I bet there's a setting in my 2nd pass that uses the PSNR for something. (like --subme 6?) I can only assume that the one which uses PSNR is higher quality, but I can't tell the difference.

[JoeBG]

Wait a second... I only get bit-identical results on the first pass.

This would say, that we can use it for a fast first pass?

[akupenguin]

Wait a second... I only get bit-identical results on the first pass. The second pass is not identical. I did a 200-frame test of it, and only about half the frames are identical. A few of the frames even use different quantizers. I bet there's a setting in my 2nd pass that uses the PSNR for something. (like --subme 6?) I can only assume that the one which uses PSNR is higher quality, but I can't tell the difference.

If so, it's a bug. RDO does include a PSNR calculation, but it's separate from the one controlled by --no-psnr.

[omion]

Oops. Looks like I made a mistake. Apparently, one of the filters that I was using in the AVS file is not repeatable. I was getting a different file for each encode, even if they both had the same settings. Once I took out that filter, all encodes were the same.