[Bug 707361] video: Add support for 64x32 tiled NV12 color format

GStreamer (bugzilla.gnome.org) bugzilla at gnome.org
Mon Dec 30 08:57:53 PST 2013


https://bugzilla.gnome.org/show_bug.cgi?id=707361
  GStreamer | gst-plugins-base | git

--- Comment #44 from Nicolas Dufresne <nicolas.dufresne at collabora.co.uk> 2013-12-30 16:57:41 UTC ---
(In reply to comment #43)
> > 
> > What name will we give to the next variant I have kept on hold ?
> > 
> > It's NV12, with tile mode "LINEAR", and tile sizes is 16x16. Stride alignment
> > should be 16. (was the reason I have put 64x32 in the name, even though not
> > very cute API whise).
> 
> V4L2 has these variants:
> 
> TM12 -> NV12MT
> VM12 -> NV12MT_16X16
> HM12 -> HM12 is like NV12MT_16X16 but contiguous planes AFAIU
> 
> No idea if these are 'official' fourccs. These fourccs don't look very
> descriptive making them very hard to remember, which is not a good property for
> things you put on caps.
> 
> Lets go for the more scriptive NV12_16L16 and NV12_64Z32 then, as you
> suggested? I put the Linear and Zigzag marker in there..

I have no ideas either, though this naming seems just right to me, and should
scale quite well.

> What we really would need is the size of the plane in the GstVideoMeta (either
> as size or as number of strides). It is needed in some places (mapping,
> copying, ...) and we always end up having to guess it based on the cropped
> height, which turns out to not work so well for tiled formats. For the next ABI
> break, I think this would be best. We could derive the number of vertical tiles
> from it, like we do now for the stride. It would also allow us to store
> multiple memory blocks per plane (interesting for RTP raw video)

Indeed, next ABI break though. Streaming tiled video through RTP (1 tile per
packet) is an interesing idea. Thanks for all this work, it is very
appreciated. I'll go through the code and do some testing when I after new
year.

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.


More information about the gstreamer-bugs mailing list