[gst-devel] RGB mimetyping in gstreamer & lcs

[Ronald Bultje]

Hey Dave,

On Tue, 2003-07-15 at 21:10, David Schleef wrote:
> How does this differentiate between x1r5g5b5_le and x1b5g5r5_le?
> Do we not support BGR formats?  It's marginal, but still useful.
> I'm not going to complain about not supporting RBG formats, though.

For the moment, these aren't supported. They are in our current system.
I must honestly confess that I don't know every single format that
exists out there. Likewise, I've never known that these two different
formats actually existed individually. That's interesting (and sounds
like a bug in endianness handling elsewhere, for that matter).

Well, anyway, yes, we should support these. We could use something like
bit offset and width for the three color components, if you want, though
that'd leave us with the same situation in which we are now, already.
I'd rather see a "quick and somewhat dirty" (dirty should not be taken
literally here) way of doing it in the approach proposed by me. It'd
definately make our life easier. And for if that matters, I don't think
many of these will exist in real life anyway. I don't think we'll ever
encounter GBR video, really.

> Currently, I'm interested in exploring how well a text string
> describing the bit order would work, similar to the string above.
> The string "x1r5g5b5" would indicate 16-bit-depth, 15-bit-per-pixel
> xRRRRRGG GGGBBBBB.

Hm, yeah, but these "Gop-strings" have to be parsed, and that sucks. I'm
hoping we can find something simpler...

> FWIW, I'm not entirely comfortable with the current fourcc system
> for YUV for this reason.  There's no fourcc for planar non-subsampled
> 8-bit YUV, or for any >8-bit YUV.  This is a problem.  Do we try to
> register new fourccs, or just squat?

The thing is, we don't have to support it. We only have to support the
formats that we'll encounter in video processing. And since there don't
exist any fourccs for these formats, apparently they're not being used
in real-life.

Or we could make up our own fourccs. As long as we document it.

Ronald

Oh, PS, I'm definately not saying that we're limited to what we've
documented. I'm just saying we should document what we're using
ourselves, that makes it clear to both us and external people. :).

-- 
Ronald Bultje <rbultje at ronald.bitfreak.net>