[FFmpeg-devel] [PATCH]Fix pnm encoding with bpc set

[Carl Eugen Hoyos]

Christophe Gisquet <christophe.gisquet <at> gmail.com> writes:

> - ffmpeg's and yours (if I'm not mistaken): the signal 
> bitdepth should always match the colorspace one, and 
> bits_per_raw_sample should indicate the original bitdepth

I believe that my opinion is mostly irrelevant 
but that FFmpeg's opinion is the only one that 
counts (until you send a patch to change the 
current meaning but that implies a major version 
bump and as said, I would like to understand 
the usecase).

> I believe rescaling in the decoder, like pnmdec does, 
> is wrong.

Please provide a sample.

> > The sample is broken, you cannot use it to test
> > anything else but ticket #2966.
> 
> It isn't, it decodes fine,

No.

> the pixels are just marked differently.

The pixels have an incorrect value, this is what 
ticket #2966 is about.
Fortunately, the ticket is reproducible with ffplay.

[...]

> > I believe it is very useful and works as expected,
> > the only issue I found today is that it gets lost
> > on endian conversion in libswscale (that should be
> > lossless).

Sorry, I should not have mentioned this, it is 
absolutely unrelated to this issue.
(That pnmenc is currently broken.)

[...]

> > Because of ticket #2966.
> 
> To fix according to your view, it needs rescaling. 
> Are you going to implement pnmdec's one?

Only if you provide a broken sample;-)

> > The values have the right dynamics except for the
> > sample from ticket #2966.
> 
> There are several codecs that could output values not ok.

Please provide samples, I thought that the issue 
only exists in mjpeg / ljpeg.

Carl Eugen