Handbrake settings - question

[JackForrester]

Hello you all,

This is my first post here. I'm sorry if this was discussed already in another thread. If so, please someone point me in the right direction and I'll delete this post. 
I've been using Handbrake to make my mkv files smaller and keep a good quality but I can't seem to understand how the compression in that program works. Here's an example: I added a 12 GB mkv file to Handbrake, compressed the file using x265 and at the end, the resulting file was 1.22 GB! I was like "no way, that's way too compressed, quality is probably shit now". Thing is, I couldn't tell the files apart, the PQ on the 1.22 Gb file was terrific (well, at least in my computer, I'll probably notice the difference when I watch it in a 4K TV).
So I decided to use the exact same settings on another 12 GB mkv file (same size but completely different file). And when the process was over, the file was 5 GB. Ok, not bad, but how come one file results in 1.22 GB and the other in 5 GB if I used the exact same settings and they're the same file size? Btw, the file that resulted in 1.22 GB is a much longer movie. So it doesn't make sense. 

PS: Yes, I'm a complete newbie to this.

[Chewtobacca]

With CRF encoding, the output size depends on how well your source compresses.  Sources that are difficult to compress (noisier, more grain, etc) require more bitrate and result in bigger file-sizes (as long as you keep the CRF value and other settings the same) than sources that are easier to compress.  Despite what you might read online, there's no one-size-fits-all bitrate that delivers a certain quality at a given resolution.

If you want bigger file-sizes, lower the CRF value, e.g., from 23 to 18.  But the best thing to do is encode a clip multiple times using a range of CRF values and find the one that gives you the quality you want at a size that you find acceptable.  Then use that value for all your CRF encodings.

[JackForrester]

Thanks for your reply.
Yeah, the CRF value I'm currently using satisfies me in terms of PQ/file size proportion. I was just curious why there's such a striking difference between the output file sizes if the original files have pretty much the same size.

[deleted user]

As Chewtobacca says, it's heavily dependent on the source. Grain/noise is notoriously difficult to compress, and a lot of motion also increases the necessary bitrate.