I have not been able to follow CinGG lately, but it seems OK now. I see many long threads and several patches from Andrew. Could you please point me to which patches to put in and what to test?
вт, 8 окт. 2024 г., 12:01 Andrea paz via Cin <[email protected]>:
I have not been able to follow CinGG lately, but it seems OK now. I see many long threads and several patches from Andrew. Could you please point me to which patches to put in and what to test?
I was mostly looking at qsv (intel-specific) encoding, and related mediacodec (but mediacodec only relevant/usable in termux). So, not sure if this part applicable for you on AMD. Another part was about ffmpeg-7.1/x265 4.0 , there was one failing filter (F_scale) that can be fixed by altering ffmpeg sources. I am not sure I want 7.1 without some range conversion fixes, but I am not sure if they ever find their way into 7.1 branch ... so, in short unless you want to see how ffmpeg 7.1 performs I do mot think there is much else to test yet? --
Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin
OK, thanks. Il mar 8 ott 2024, 11:10 Andrew Randrianasulu <[email protected]> ha scritto:
вт, 8 окт. 2024 г., 12:01 Andrea paz via Cin <[email protected]>:
I have not been able to follow CinGG lately, but it seems OK now. I see many long threads and several patches from Andrew. Could you please point me to which patches to put in and what to test?
I was mostly looking at qsv (intel-specific) encoding, and related mediacodec (but mediacodec only relevant/usable in termux).
So, not sure if this part applicable for you on AMD.
Another part was about ffmpeg-7.1/x265 4.0 , there was one failing filter (F_scale) that can be fixed by altering ffmpeg sources. I am not sure I want 7.1 without some range conversion fixes, but I am not sure if they ever find their way into 7.1 branch ...
so, in short unless you want to see how ffmpeg 7.1 performs I do mot think there is much else to test yet?
--
Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin
Den 08.10.2024 11:09, skrev Andrew Randrianasulu via Cin:
вт, 8 окт. 2024 г., 12:01 Andrea paz via Cin <[email protected]>:
I have not been able to follow CinGG lately, but it seems OK now. I see many long threads and several patches from Andrew. Could you please point me to which patches to put in and what to test?
I was mostly looking at qsv (intel-specific) encoding, and related mediacodec (but mediacodec only relevant/usable in termux).
So, not sure if this part applicable for you on AMD.
To verify if the last two patches I applied 28/9-24 via: Re: [Cin] so, it looks like qsv integration a bit harder than just adding few lines .... 0001-Add-nv12-as-def.-pixformat-for-qsv-profiles-clarify-.patch 0002-Modify-remaining-_qsv-orofiles-to-use-nv12-fix-typo-.patch caused some of my later testing issues "mismatch audio/video file format", I wonder if it is easy roll back/"unpatch" to my system status before patching; possibly how to do it?
Another part was about ffmpeg-7.1/x265 4.0 , there was one failing filter (F_scale) that can be fixed by altering ffmpeg sources. I am not sure I want 7.1 without some range conversion fixes, but I am not sure if they ever find their way into 7.1 branch ...
so, in short unless you want to see how ffmpeg 7.1 performs I do mot think there is much else to test yet?
Although basic ffmpeg 7.1 has been available for Slowroll since a little week (from obs:multimedia:libs) I thought I would wait a bit longer with testing until the 7.1 series becomes full codec enabled from Packman.
-- Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin
вт, 8 окт. 2024 г., 15:46 Terje J. Hanssen <[email protected]>:
Den 08.10.2024 11:09, skrev Andrew Randrianasulu via Cin:
вт, 8 окт. 2024 г., 12:01 Andrea paz via Cin <[email protected]>:
I have not been able to follow CinGG lately, but it seems OK now. I see many long threads and several patches from Andrew. Could you please point me to which patches to put in and what to test?
I was mostly looking at qsv (intel-specific) encoding, and related mediacodec (but mediacodec only relevant/usable in termux).
So, not sure if this part applicable for you on AMD.
To verify if the last two patches I applied 28/9-24 via: Re: [Cin] so, it looks like qsv integration a bit harder than just adding few lines ....
0001-Add-nv12-as-def.-pixformat-for-qsv-profiles-clarify-.patch 0002-Modify-remaining-_qsv-orofiles-to-use-nv12-fix-typo-.patch
caused some of my later testing issues "mismatch audio/video file format", I wonder if it is easy roll back/"unpatch" to my system status before patching; possibly how to do it?
use git log in cingg source directory. you will see many lines like commit 9bc167893466062f9b5f59be6ebc68c96557cbf7 (HEAD -> master) Author: Andrew Randrianasulu <[email protected]> Date: Tue Oct 8 16:04:04 2024 +0300 you can select long string after commit word, just before two recently applied commits. then 'q' from git log's pager (usually less) and try git reset --hard long-string-you-just copied verify with 'git log' that two last commits are gone
Another part was about ffmpeg-7.1/x265 4.0 , there was one failing filter (F_scale) that can be fixed by altering ffmpeg sources. I am not sure I want 7.1 without some range conversion fixes, but I am not sure if they ever find their way into 7.1 branch ...
so, in short unless you want to see how ffmpeg 7.1 performs I do mot think there is much else to test yet?
Although basic ffmpeg 7.1 has been available for Slowroll since a little week (from obs:multimedia:libs) I thought I would wait a bit longer with testing until the 7.1 series becomes full codec enabled from Packman.
--
Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin
So glad Andrea asked this question as I was wondering if I missed anything. I will check in x265 version 4.0 along with the patch Andrew provided as ffmpeg-7.0.patch6 to fix it tomorrow. Also, a question for Andrew about F_scale: if there is a specific parameter setting that makes this not fail, it can be added to cinelerra-5.1/ffmpeg/plugin.opts. Currently there is a line "scale 200:100" in that file that sets width and height defaults. So if the parameter is in 7.0 as well as 7.1, we can set that now so it does not fail later. And another question for Andrew, I thought in one of your emails you mentioned vf_filter or something like that that also might need to be patched. But I can not find that anymore. On Tue, Oct 8, 2024 at 3:10 AM Andrew Randrianasulu via Cin < [email protected]> wrote:
вт, 8 окт. 2024 г., 12:01 Andrea paz via Cin <[email protected]>:
I have not been able to follow CinGG lately, but it seems OK now. I see many long threads and several patches from Andrew. Could you please point me to which patches to put in and what to test?
I was mostly looking at qsv (intel-specific) encoding, and related mediacodec (but mediacodec only relevant/usable in termux).
So, not sure if this part applicable for you on AMD.
Another part was about ffmpeg-7.1/x265 4.0 , there was one failing filter (F_scale) that can be fixed by altering ffmpeg sources. I am not sure I want 7.1 without some range conversion fixes, but I am not sure if they ever find their way into 7.1 branch ...
so, in short unless you want to see how ffmpeg 7.1 performs I do mot think there is much else to test yet?
--
Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin
-- Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin
ср, 9 окт. 2024 г., 01:42 Phyllis Smith <[email protected]>:
So glad Andrea asked this question as I was wondering if I missed anything. I will check in x265 version 4.0 along with the patch Andrew provided as ffmpeg-7.0.patch6 to fix it tomorrow.
Also, a question for Andrew about F_scale: if there is a specific parameter setting that makes this not fail, it can be added to cinelerra-5.1/ffmpeg/plugin.opts. Currently there is a line "scale 200:100" in that file that sets width and height defaults. So if the parameter is in 7.0 as well as 7.1, we can set that now so it does not fail later.
unfortunately, ffmpeg's vf_scale was improved so now it takes dynamic input, and by default cingg filters out such filters https://git.ffmpeg.org/gitweb/ffmpeg.git/blob/HEAD:/libavfilter/vf_scale.c#l... .flags = AVFILTER_FLAG_DYNAMIC_INPUTS, https://git.cinelerra-gg.org/git/?p=goodguy/cinelerra.git;a=blob;f=cinelerra... 1258 int flag_mask = AVFILTER_FLAG_DYNAMIC_INPUTS | AVFILTER_FLAG_DYNAMIC_OUTPUTS; 1259 if( filter->flags & flag_mask ) return AVERROR(EPERM); I commented out (//) .flags line in ffmpeg, but may be we can comment out those two lines in our pluginfclient.C and see if it still works?
And another question for Andrew, I thought in one of your emails you mentioned vf_filter or something like that that also might need to be patched. But I can not find that anymore.
On Tue, Oct 8, 2024 at 3:10 AM Andrew Randrianasulu via Cin < [email protected]> wrote:
вт, 8 окт. 2024 г., 12:01 Andrea paz via Cin <[email protected]
:
I have not been able to follow CinGG lately, but it seems OK now. I see many long threads and several patches from Andrew. Could you please point me to which patches to put in and what to test?
I was mostly looking at qsv (intel-specific) encoding, and related mediacodec (but mediacodec only relevant/usable in termux).
So, not sure if this part applicable for you on AMD.
Another part was about ffmpeg-7.1/x265 4.0 , there was one failing filter (F_scale) that can be fixed by altering ffmpeg sources. I am not sure I want 7.1 without some range conversion fixes, but I am not sure if they ever find their way into 7.1 branch ...
so, in short unless you want to see how ffmpeg 7.1 performs I do mot think there is much else to test yet?
--
Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin
-- Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin
participants (4)
-
Andrea paz -
Andrew Randrianasulu -
Phyllis Smith -
Terje J. Hanssen