x265 encode for HW player

[nevcairiel]

I agree with the frame dup, drop it, no need to make the video practically VFR for a very tiny gain only, and potentially cause issues with muxers and/or players.

[jpsdr]

Ok, thanks for the info, just hope in the future they fix this preset bug, honestly....

[jpsdr]

Seriously........ this is bugland !!!!!!!!!!!!!!!!!

Some parts of the video is totaly screwed up, and i have no idea what setting caused it....

The command line was (tuned animation) :

@echo off

SET E_SRC=%8%1.avs
SET E_DST=%5%1.hevc
SET E_DOVI=%8%1
SET CHAPTERS=%8%7
SET STAT_FILE=%8%1.stats
SET LOG_FILE_1=%8%1_log_1.txt
SET LOG_FILE_2=%8%1_log_2.txt
SET LOG_FILE_3=%8%1_log_3.txt
SET BITRATE=%2
SET TUNING=%6
SET MCLL=%3
SET MDISPLAY=%4

if %6==film goto :NOTUNE
if %6==none goto :NOTUNE
if %6==aucun goto :NOTUNE

x265_x64 --preset slower --tune %TUNING% --vbv-maxrate 90000 --vbv-bufsize 70000 --bitrate %BITRATE% --stats %STAT_FILE% --level 5.1 --profile main10 --high-tier --level-idc 51 --fades --aq-mode 3 --hevc-aq --weightb --rc-lookahead 48 --rect --amp --b-intra --no-sao --scenecut-aware-qp 3 --analyze-src-pics --opt-ref-list-length-pps --opt-qp-pps --multi-pass-opt-distortion --multi-pass-opt-rps --colorprim bt2020 --transfer smpte2084 --colormatrix bt2020nc --range limited --chromaloc 2 -D 10 --max-cll %MCLL% --master-display %MDISPLAY% --hdr10-opt --qpfile %CHAPTERS% --input %E_SRC% --pass 1 -o NUL 2> %LOG_FILE_1%
x265_x64 --preset slower --tune %TUNING% --vbv-maxrate 90000 --vbv-bufsize 70000 --bitrate %BITRATE% --stats %STAT_FILE% --level 5.1 --profile main10 --high-tier --level-idc 51 --fades --aq-mode 3 --hevc-aq --weightb --rc-lookahead 48 --rect --amp --b-intra --no-sao --scenecut-aware-qp 3 --analyze-src-pics --opt-ref-list-length-pps --opt-qp-pps --multi-pass-opt-distortion --multi-pass-opt-rps --colorprim bt2020 --transfer smpte2084 --colormatrix bt2020nc --range limited --chromaloc 2 -D 10 --max-cll %MCLL% --master-display %MDISPLAY% --hdr10-opt --qpfile %CHAPTERS% --input %E_SRC% --pass 3 -o NUL 2> %LOG_FILE_2%
x265_x64 --preset slower --tune %TUNING% --vbv-maxrate 90000 --vbv-bufsize 70000 --bitrate %BITRATE% --stats %STAT_FILE% --level 5.1 --profile main10 --high-tier --level-idc 51 --fades --aq-mode 3 --hevc-aq --weightb --rc-lookahead 48 --rect --amp --b-intra --no-sao --scenecut-aware-qp 3 --analyze-src-pics --opt-ref-list-length-pps --opt-qp-pps --multi-pass-opt-distortion --multi-pass-opt-rps --colorprim bt2020 --transfer smpte2084 --colormatrix bt2020nc --range limited --chromaloc 2 -D 10 --max-cll %MCLL% --master-display %MDISPLAY% --hdr10-opt --dolby-vision-profile 81 --dolby-vision-rpu %E_DOVI%.bin --qpfile %CHAPTERS% --input %E_SRC% --pass 2 -o %E_DST% 2> %LOG_FILE_3%
goto :FIN

:NOTUNE
x265_x64 --preset slower --vbv-maxrate 90000 --vbv-bufsize 70000 --bitrate %BITRATE% --stats %STAT_FILE% --level 5.1 --profile main10 --high-tier --level-idc 51 --fades --aq-mode 3 --hevc-aq --weightb --rc-lookahead 48 --rect --amp --b-intra --no-sao --scenecut-aware-qp 3 --analyze-src-pics --opt-ref-list-length-pps --opt-qp-pps --multi-pass-opt-distortion --multi-pass-opt-rps --colorprim bt2020 --transfer smpte2084 --colormatrix bt2020nc --range limited --chromaloc 2 -D 10 --max-cll %MCLL% --master-display %MDISPLAY% --hdr10-opt --qpfile %CHAPTERS% --input %E_SRC% --pass 1 -o NUL 2> %LOG_FILE_1%
x265_x64 --preset slower --vbv-maxrate 90000 --vbv-bufsize 70000 --bitrate %BITRATE% --stats %STAT_FILE% --level 5.1 --profile main10 --high-tier --level-idc 51 --fades --aq-mode 3 --hevc-aq --weightb --rc-lookahead 48 --rect --amp --b-intra --no-sao --scenecut-aware-qp 3 --analyze-src-pics --opt-ref-list-length-pps --opt-qp-pps --multi-pass-opt-distortion --multi-pass-opt-rps --colorprim bt2020 --transfer smpte2084 --colormatrix bt2020nc --range limited --chromaloc 2 -D 10 --max-cll %MCLL% --master-display %MDISPLAY% --hdr10-opt --qpfile %CHAPTERS% --input %E_SRC% --pass 3 -o NUL 2> %LOG_FILE_2%
x265_x64 --preset slower --vbv-maxrate 90000 --vbv-bufsize 70000 --bitrate %BITRATE% --stats %STAT_FILE% --level 5.1 --profile main10 --high-tier --level-idc 51 --fades --aq-mode 3 --hevc-aq --weightb --rc-lookahead 48 --rect --amp --b-intra --no-sao --scenecut-aware-qp 3 --analyze-src-pics --opt-ref-list-length-pps --opt-qp-pps --multi-pass-opt-distortion --multi-pass-opt-rps --colorprim bt2020 --transfer smpte2084 --colormatrix bt2020nc --range limited --chromaloc 2 -D 10 --max-cll %MCLL% --master-display %MDISPLAY% --hdr10-opt --dolby-vision-profile 81 --dolby-vision-rpu %E_DOVI%.bin --qpfile %CHAPTERS% --input %E_SRC% --pass 2 -o %E_DST% 2> %LOG_FILE_3%

:FIN

EDIT :
Noticed the video is screwed up at a place i put an I frame in the chapter file.
EDIT2 :
Can reproduce with a small 200 frames centered where the issue occurs.

[jpsdr]

Fortunately i was able to reproduce with a small video, allowing me to remove options one at a time until i found the guilty.

DO NOT, NEVER USE --multi-pass-opt-rps !!!!!!!!!
It's hell bugged, and it will screw up the result video ! I'm not talking result look bad, i'm talking unplayable blocked pixelised totaly screwed up result stream.

[rwill]

Originally Posted by jpsdr

Fortunately i was able to reproduce with a small video, allowing me to remove options one at a time until i found the guilty.

DO NOT, NEVER USE --multi-pass-opt-rps !!!!!!!!!
It's hell bugged, and it will screw up the result video ! I'm not talking result look bad, i'm talking unplayable blocked pixelised totaly screwed up result stream.

May I ask why you are also using "--analyze-src-pics" ?

[excellentswordfight]

Originally Posted by jpsdr

--aq-mode 3 --hevc-aq

Why would you set both mode 3 and hevc-aq? hevc-aq will disable other aq modes, and is afaik still experimental.

[jpsdr]

I've spend a lot of time reading the 3.5 doc of x265 and all the options, and activate everything which seems "good" for quality.

To answer the questions :

@rwill
The doc said :

Enable motion estimation with source frame pixels, in this mode, motion estimation can be computed independently.

From the description, it seems it's a good thing.
From your experience, it's not ?

@excellentswordfight
The doc said :

Enable adaptive quantization It scales the quantization step size according to the spatial activity of one coding unit relative to frame average spatial activity.
This AQ method utilizes the minimum variance of sub-unit in each coding unit to represent the spatial complexity of the coding unit.

There is nothing saying it's experimental, also nothing saying it disables the others AQ mode (even if i've noticed it later when seeing the log files).
According the description, it also seems a good option to activate.
From your experience, it's not ?

[rwill]

Originally Posted by jpsdr

@rwill
The doc said :

Enable motion estimation with source frame pixels, in this mode, motion estimation can be computed independently.

From the description, it seems it's a good thing.
From your experience, it's not ?

Well I have no experience with that but in theory its better to use the reconstructed pixels because these are actually available on the decoder side and are the only thing a prediction can be formed from. I don't think doing ME on source pixels as reference is any good because inter prediction is formed from a reference picture anyway ..