one pass CRF with multiple video

[sosotiit]

is there a way to have the CRF prediction algorithm evaluate all the video and determine one CRF for all of them?

have two movies with a menu using multiavchd, first got encoded at crf 14 the second at 21. Probably allocation proportionally to original file size??

- [20:15:38] Performing CRF Prediction...
- Analyzing 11.00 30.50 19.92 15.58 14.52 [14.40]
...
- [22:56:53] Performing CRF Prediction...
- Analyzing 20.32 35.16 24.25 21.84 21.61 [21.60]

or is there any way to run the prediction test separately , then I could force the encoding at a certain crf
or have a quick first pass? as 100% precision on size is not my key objective?

edit- - -
in order to better understand I did two encode of these movies separately with FORCE_ENCODE=1, FIXED_CRF=18 and the first movie got reduced to 4.87GB!!! (from 25.5GB) and the second was only reduced to 19.3GB.





tia, soso

[jdobbs]

It definitely allocates according to the original size. The only way to give them equal CRF is how you did it -- and you can see the result. You're better off letting BD-RB do it proportionally. That way the encode is considering the complexity of the source and allocating correctly.

[Asmodian]

I personally would rather have the video at a specific quality and not make the more complex ones look worse and the less complex ones waste bits on details I cannot see. I would just use crf 18 for both, is there a reason you cannot use the results from your test?

I don't understand analyzing videos to pick a crf, isn't the point of crf to set quality? If crf 21 is good enough just use crf 21 for everything. If crf 21 is not good enough do you want some videos "not good enough" just because they required more bits to encode well? I have not had large variations in quality using the same crf (and other settings) on different sources so unless you want variations in quality I wouldn't change crf between videos.

[jdobbs]

I personally would rather have the video at a specific quality and not make the more complex ones look worse and the less complex ones waste bits on details I cannot see. I would just use crf 18 for both, is there a reason you cannot use the results from your test?

I don't understand analyzing videos to pick a crf, isn't the point of crf to set quality? If crf 21 is good enough just use crf 21 for everything. If crf 21 is not good enough do you want some videos "not good enough" just because they required more bits to encode well? I have not had large variations in quality using the same crf (and other settings) on different sources so unless you want variations in quality I wouldn't change crf between videos. I think you hit the nail on the head. You "don't understand". The analysis is to pick the best CRF that will fit in the required space. It you set a CRF and it won't fit on the disc -- then what's the point??? You'd just have to do it again.