variables of x264 encodes filesizes

[skx7]

Hi all

I am backing up my whole DVD cartoon collection. While encoding all with same x264 settings I noticed that older cartoons (made before 1980) consistently end up with more substantial filesizes compared to more recent animation movies. What can explain this difference in encoding efficiency? Just wondering what is triggering these differences to maybe identify what can be changed to reduce the filesizes without substantial quality loss.

Thanks

[kalehrl]

They probably contain more noise which is hard to compress than the more recent ones.

[Forteen88]

You might want to degrain them to save bitrate.
But don't degrain too much, since that makes the image-quality suffer.

[benwaggoner]

Using x264's --nr is actually pretty good for this scenario, since it takes out exactly the noise that makes it harder to compress. Which, for anime, is pretty much the noise that wasn't in the original cels.

[zub35]

Use nr has meaning only in the range of 100-300. The values are larger, not rational, compared to increasing crf.
for example, crf=20 nr=500 will be worse than crf=20.5 nr=0, with the same bitrate

In some files, it is better to strengthen psy, keeping roughly the same bitrate crf=20 vs crf=21 psy=1.2:0.2
In this case, keeping the bitrate identical, we make the codec focus more on noise and fine detail changing (not removing) it for better compression.

You can also apply them in a bundle crf=20 vs crf=21 nr=300 psy=1.2:0.2
This will help to reduce the bitrate, without significant deterioration of perception.

[benwaggoner]

Originally Posted by zub35

Use nr has meaning only in the range of 100-300. The values are larger, not rational, compared to increasing crf.
for example, crf=20 nr=500 will be worse than crf=20.5 nr=0, with the same bitrate

But crf=20 nr=500 will almost always be a lower bitrate than crf=20.5 nr=0. That nr=500 will take some of hard-to-compress coefficients, like an adaptive deadzone.