[FFmpeg-user] remap filter, unmapped pixels

[Michael Koch]

Am 20.02.2020 um 12:49 schrieb Paul B Mahol:
> On 2/20/20, Michael Koch <astroelectronic at t-online.de> wrote:
>> Am 20.02.2020 um 11:07 schrieb Paul B Mahol:
>>> On 2/20/20, Michael Koch <astroelectronic at t-online.de> wrote:
>>>> I'd like to understand how the new "fill" option in the remap filter
>>>> works.
>>>> What is an "unmapped pixel"? How is it possible that pixels are
>>>> unmapped, if the xmap and ymap files have the same size as the output
>>>> image? Is a pixel "unmapped" if it has invalid coordinates (negative or
>>>> too big) in the xmap or ymap files?
>>> Previously all unmapped pixels were set to black for rgba or dark
>>> green for y420p.
>>> It is very obvious to understand even for every little kid in the block.
>> Finally I got it working. A pixel can be declared as "unmapped" by
>> setting the values to 65535 in the xmap and ymap files. However using a
>> negative value (for example -1) doesn't work. In this case the pixel is
>> mapped to the top left pixel of the input image.
>> This behaviour might be obvious for little kids, but not for me. Should
>> be added to documentation.
> Nope. Documentation does not need trivial information.
> Obviously by having even limited programming experience one would
> found out that remaps are unsigned short, thus can not be negative.

I that case I did make the PGM file with C# code and some coordinates 
were set to -1. Obviously I shouldn't have done that because the PGM 
specification doesn't allow negative values. But FFmpeg did not report 
any errors when reading this file. May be a warning should be added if 
negative values are found.

Michael