I thought I knew what I was doing with Handbrake encoding, but now I don’t know what to think…

UPDATE: This StackExchange question might explain what I am seeing with these tests. tl;dr: CRF is not true constant quality.

I always thought video encoding followed the simple formula of Quality, Speed, Filesize: Pick Two. So I would choose the best, reasonable quality setting for the video and encode as slowly as reasonable to get the smallest filesize. However, after some testing with x265 I have to throw all that thinking out the window.

I recently bought the Daredevil Season 1 blu-ray and decided to rip them with x265 as my Plex server is more powerful for transcoding and more of my client devices can do x265 natively.

I used Handbrake for the tests, and I started with the built-in preset of H.265 MKV 1080p30 and then adjusted it closer to my normal settings:

  • Removed the de-interlace filters.
  • Removed audio tracks for the tests.
  • Set the framerate to Same as Source/variable.
  • The preset also included strong-intra-smoothing=0:rect=0:aq-mode=1, so I kept those settings.

Then I just re-encoded the same chapter at different speeds.

Here are the results for Daredevil S01E01 Chapter 3 (11 minutes):

x265 testing results

I have no idea how or why increasing the encoding speed consistently results in the smaller filesize, but it does.