RFC: New format V4L2_PIX_FMT_Y16_BE ?

RFC: New format V4L2_PIX_FMT_Y16_BE ?

[Ricardo Ribalda Delgado]

Hello

While debugging a v4l2<->gstreamer interaction problem, I have
realized that we were implementing Y16 wrong in our cameras :S. What
we were implementing was a big endian version of the format.

If I want to add support for our Y16_BE format to the kernel, would it
be enough to change videodev2.h, add the documentation for the format
and update libv4lconvert?


Regards!


-- 
Ricardo Ribalda

Re: RFC: New format V4L2_PIX_FMT_Y16_BE ?

[Hans Verkuil]

On 03/26/2015 09:23 AM, Ricardo Ribalda Delgado wrote:
> Hello
> 
> While debugging a v4l2<->gstreamer interaction problem, I have
> realized that we were implementing Y16 wrong in our cameras :S. What
> we were implementing was a big endian version of the format.
> 
> If I want to add support for our Y16_BE format to the kernel, would it
> be enough to change videodev2.h, add the documentation for the format
> and update libv4lconvert?

Yes, but you might want to wait 1-2 weeks: I'm going to make a patch
that moves the ENUMFMT description into the v4l2 core. So you want to be on
top of that patch. I posted an RFC patch for that earlier (last week I
think).

Regards,

	Hans

Re: RFC: New format V4L2_PIX_FMT_Y16_BE ?

[Ricardo Ribalda Delgado]

Hi Hans

On Thu, Mar 26, 2015 at 5:46 PM, Hans Verkuil <hverkuil@xs4all.nl> wrote:

> Yes, but you might want to wait 1-2 weeks: I'm going to make a patch
> that moves the ENUMFMT description into the v4l2 core. So you want to be on
> top of that patch. I posted an RFC patch for that earlier (last week I
> think).

Absolutely no hurry.

My main worry is now this patch:

libv4lconvert: Fix support for Y16 pixel format
https://patchwork.linuxtv.org/patch/28989/

That fixes the implementation in v4lconvert. I introduced the original
bug and I don't want that anybody uses that library as a reference of
how the format should be.



Thanks!!



-- 
Ricardo Ribalda