[FFmpeg-devel] [PATCH] pgssubdec: fix subpicture output colorspace and range

[Jan Ekstrom]

On Sat, Apr 23, 2016 at 11:09 AM, Carl Eugen Hoyos <cehoyos at ag.or.at> wrote:
> Please mention the relevant ticket in the commit message.

OK, will do.

>
>> +    int hdtv;
>
> Please rename to sdtv so you can remove the assignment from
> init_decoder().
>

What would that make of the default value? The current design lets me
test for C true/false since the default is overridden at init() to -1
and the HD content would have 1 set. I seem to remember that in stack
various values are set to zero, but is the context always on stack?

>
> ctx->sdtv = h <=576 (or add braces).
>

OK, will set the value according to the response to my previous question.

>> +#define YUV_TO_RGB1_CCIR_BT709(cb1, cr1)\
>
> Just being curious: Did you test how much the values were
> off without this hunk?
>

The usual amount of BT.601 vs BT.709. Granted, many subpictures don't
contain much if any color, but since the values are not really limited
I'd say it's only fair to compare it to a standard BT.709/BT.601
rendering difference. If you haven't experienced such difference yet,
you can test it with these following files that contain the exact same
YCbCr values, but different colorimetry metadata in the bit stream
(601 and 709 accordingly):
http://cccp-project.net/beta/test_files/renderer_test_with_bt601.mkv
http://cccp-project.net/beta/test_files/renderer_test_with_bt709.mkv

Jan