TomArrow
15th January 2019, 03:07
I was wondering if there was a way to load an h264 file in such a way that not the image is displayed, but rather the quantization multiplier of each block is displayed visually, so that you could do the following:
A.) Optimize bit allocation if you are an encoder
B.) Combine two different encodes of the same source for maximum quality
My main desire is B. Imagine you have the same source, but two different encodes. Of course you can just average them out or Median if you have more than 2.
But that isn't really taking the potential quality to the limit I think. If you could display the bit allocation/quantization multiplier visually for each encode, you could create masks out of that information with a bit of extra processing and then simply use the better looking blocks with more allocated bits from each encode, combining both encodes for maximum quality.
Of course I'm not sure how well that would work in encodes made using strong psy-rd or other psychovisual algorithms, but I think it's worth a try nonetheless.
A.) Optimize bit allocation if you are an encoder
B.) Combine two different encodes of the same source for maximum quality
My main desire is B. Imagine you have the same source, but two different encodes. Of course you can just average them out or Median if you have more than 2.
But that isn't really taking the potential quality to the limit I think. If you could display the bit allocation/quantization multiplier visually for each encode, you could create masks out of that information with a bit of extra processing and then simply use the better looking blocks with more allocated bits from each encode, combining both encodes for maximum quality.
Of course I'm not sure how well that would work in encodes made using strong psy-rd or other psychovisual algorithms, but I think it's worth a try nonetheless.