Kishuku Gakkou no Juliet [BD 1080p HEVC_10 AAC]
After further testing my encodes will continue with limit-sao. I won't be getting into filtering as of now (you can alwas run your own shaders/AVIsynth scripts while watching). no-sao above crf=16 has ringing artifacts and doesn't really increase sharpness compared to limit-sao. In fact sao actually looked visually the best at crf=20 while limit-sao looked best for crf=18. I will continue using limit-sao as a balanced medium. Test clips used are BDMV of NGNL Zero (high motion scene, grainy/noisy scene, flat dark scene) and Kimi no na wa.
Video Format: HEVC 10 bit (crf=20, preset=slow w/ bframes=8:limit-sao:psy-rd=1:aq-mode=3 override)
Audio Format: AAC (FDK AAC; VBR Q5 ~220kbps)
Subtitle: ASS (HorribleSubs)
-
Kishuku Gakkou no Juliet
- Kishuku Gakkou no Juliet 01.mkv (600.2 MiB)
- Kishuku Gakkou no Juliet 02.mkv (593.4 MiB)
- Kishuku Gakkou no Juliet 03.mkv (466.7 MiB)
- Kishuku Gakkou no Juliet 04.mkv (419.4 MiB)
- Kishuku Gakkou no Juliet 05.mkv (566.4 MiB)
- Kishuku Gakkou no Juliet 06.mkv (701.8 MiB)
- Kishuku Gakkou no Juliet 07.mkv (407.4 MiB)
- Kishuku Gakkou no Juliet 08.mkv (460.9 MiB)
- Kishuku Gakkou no Juliet 09.mkv (455.1 MiB)
- Kishuku Gakkou no Juliet 10.mkv (387.0 MiB)
- Kishuku Gakkou no Juliet 11.mkv (400.7 MiB)
- Kishuku Gakkou no Juliet 12.mkv (392.6 MiB)
I used the highest quality anime mastering I could think or for my tests, as I want my encoded settings to be based on it. I actually also did some tests in normal anime and also preferred limit-sao although sao does make it more visually pleasing especially at crf=20. When I do crf=20 it's usually a re-encode or a badly mastered series.
I used to use qAAC but using an FDK custom build is musch easier for my workflow since it's built in. It's ever so slightly worse at low bitrates (tests were done at 96kbps I believe), and considering I do -q 5 which can boost as high as ~400 kbps on complex passages I consider it to be transparent to most, if not all people.
No issues here with what you decided, however, using two source (NGNL Zero and Kimi no na wa) to do your testing that differs to this one the results may not be the same more than likely and what version of Kimi no na wa you referring to. Further, turning sao on for re-encodes might be a good thing since the source comes from lossy but not so from a so called lossless source i.e. a BDMV. Well unless it's a grainy source or something.
I'm not discounting the other factors may come into play too like if there's an intent on lowering file size, or high CRF value (18 and up which you've outlined above) and so forth.
I also just noticed your AAC audio choice. Have you considered qAAC?