Posts: 53
Threads: 8
Joined: Nov 2020
I noticed a problem to set the value in the UI cell related to "Quantizer max" (x264-Quantization), I tried to restore the default profile thinking that it depended on this but there is the same behavior.
- I load the x264 hybrid default configuration
- I see that the quoted value is 81
- Change AVC Profile to (for example) None
- The value is now 51
- AVC Profile High replacement (previous value)
- The value remains at 51 and I cannot increase it even when reloading the configuration, I have to restart the program.
Posts: 10.985
Threads: 57
Joined: May 2017
04.12.2020, 19:50
(This post was last modified: 04.12.2020, 19:53 by Selur.)
That's a bug,..
atm. Hybrid uses "51 + 6 * (depth - 8)"
but it should be:
#define QP_BD_OFFSET (6*(BIT_DEPTH-8))
#define QP_MAX_SPEC (51+QP_BD_OFFSET)
#define QP_MAX (QP_MAX_SPEC+18)
#define QP_MAX_MAX (51+2*6+18)
see:
https://github.com/cybertk/x264/blob/mas...n/common.h line 62+
-> Hybrid missed the '+18' for quantizer max and the default should be 69 since the default bit depth is 8 bit
Posts: 10.985
Threads: 57
Joined: May 2017
Just stumpled over another problem,... '-ignore_editlist true' is only allowed for mov/mp4/3gp in ffmpeg.
Cu Selur
Posts: 53
Threads: 8
Joined: Nov 2020
(04.12.2020, 19:50)Selur Wrote: That's a bug,..
atm. Hybrid uses "51 + 6 * (depth - 8)"
but it should be:
#define QP_BD_OFFSET (6*(BIT_DEPTH-8))
#define QP_MAX_SPEC (51+QP_BD_OFFSET)
#define QP_MAX (QP_MAX_SPEC+18)
#define QP_MAX_MAX (51+2*6+18)
-> Hybrid missed the '+18' for quantizer max and the default should be 69 since the default bit depth is 8 bit
So the max settable should be 81 as i previously set (without change AVC profile)?
(I send you a PM)
Posts: 10.985
Threads: 57
Joined: May 2017
Quote:So the max settable should be 81 as i previously set (without change AVC profile)?
No.
The maximum quantizer you can use as quantizer is 51+(6*(BIT_DEPTH-8)).
The quantizer range allowed for the encoder is 51+(6*(BIT_DEPTH-8))+18.
-> Quantizer range max can only be 81 for 10bit bit depth.
Cu Selur
Posts: 53
Threads: 8
Joined: Nov 2020
Nevermind, my bad, now it works
Posts: 10.985
Threads: 57
Joined: May 2017
Looking at the debug output you selected '... (2nd pass)', which then isn't started since the stats file specified doesn't exist.
-> this is the correct behavior.
My guess is you wanted to use '.. (2pass)',..
Cu Selur
Posts: 53
Threads: 8
Joined: Nov 2020
(04.12.2020, 21:19)Selur Wrote: Looking at the debug output you selected '... (2nd pass)', which then isn't started since the stats file specified doesn't exist.
-> this is the correct behavior.
My guess is you wanted to use '.. (2pass)',..
Cu Selur
Yes, sorry, I edited after i realized (and I even selected 3 times the same wrong option
)
