I was driving myself insane trying to replicate the quality I got from a recent encode. To save time I only encoded a small chapter to do the comparison.
No matter what I did, the quality was always worse. Even with placebo-style settings that were theoretically better than the comparison's settings it turned out worse.
Frustrated, thinking I would never get these settings back again, I decided to encode the full-length file again - and the quality was way better, matching the comparison encode.
Why would this be happening?
Frame from the full-length encode:
https://ibb.co/QDNJWJn
Frame from the chapter encode (same exact settings):
https://ibb.co/vXRMMXJ
Quote:Why would this be happening?
Bit rate control didn't have enough bit rate to spare.
Wild gueses, since you didn't share encoder and the settings you use:
Use 1pass crf (or similar) in no case use 1pass abr/cbr/vbr. In 2pass encoding this could happen if the encoder would be able to 'borrow' bit rate from another nearby scene otherwise.
Interesting. Yes it was an x265 2-pass at 2000kbps. The settings for the full-length encode were straight default when initializing Hybrid, unmodified.
I made modifications from there for the chapter encode, such as higher --rd, --me, --subme.
The chapter was 6:56 minutes which I thought would be enough to average out the bit rate.
Is there a way I can see what an output will look like when using 2-pass for a short sample, like a chapter, without doing a full-length encode?
Code:
x265 --input - --output-depth 10 --y4m --profile main10 --limit-modes --no-open-gop --opt-ref-list-length-pps --lookahead-slices 0 --pass 1 --no-slow-firstpass --bitrate 2000 --opt-qp-pps --cbqpoffs -2 --crqpoffs -2 --qpfile GENERATED_QP_FILE --limit-refs 0 --ssim-rd --psy-rd 2.50 --rdoq-level 2 --psy-rdoq 10.00 --aq-mode 0 --deblock=-1:-1 --limit-sao --no-repeat-headers --stats "C:\Users\Michael\AppData\Local\Temp\movie_generated.stats" --multi-pass-opt-analysis --multi-pass-opt-distortion --analysis-reuse-file "C:\Users\Michael\AppData\Local\Temp\movie_generated.analysis" --output NUL
x265 --input - --output-depth 10 --y4m --profile main10 --limit-modes --no-early-skip --no-open-gop --opt-ref-list-length-pps --lookahead-slices 0 --pass 2 --bitrate 2000 --opt-qp-pps --cbqpoffs -2 --crqpoffs -2 --limit-refs 0 --ssim-rd --psy-rd 2.50 --rdoq-level 2 --psy-rdoq 10.00 --aq-mode 0 --deblock=-1:-1 --limit-sao --no-repeat-headers --stats "C:\Users\Michael\AppData\Local\Temp\movie_generated.stats" --multi-pass-opt-analysis --multi-pass-opt-distortion --analysis-reuse-file "C:\Users\Michael\AppData\Local\Temp\movie_generated.analysis" --output "C:\Users\Michael\AppData\Local\Temp\movie.265"
No.
Also, your bit rate is really low for such content, so using some settings which allow for more blurring might help.
Bummer. That will take a lot of time for testing lol.
I realize the bitrate is low, but I'm also fine with the quality that comes from the first image. I intentionally picked a "difficult" action frame vs a still frame. I'm impressed with the detail that is preserved of the water that is flicking off the characters.
UHD BD reference:
https://ibb.co/b5Rxx2L
I'm also keeping a 4K encode in addition to this 1080p encode.