Hi Stephan, On Thursday 17 February 2011 18:18:38 Stephan Lachowsky wrote: > On Thu, 2011-02-17 at 08:03 -0800, Laurent Pinchart wrote: > > On Friday 28 January 2011 20:35:05 Stephan Lachowsky wrote: > > > Parse the UVC 1.0 and UVC 1.1 VS_FORMAT_MPEG2TS descriptors. > > > This a stream based format, so we generate a dummy frame descriptor > > > with a dummy frame interval range. > > > > Thanks for the patch, and sorry for the late reply. > > No worries, just glad to have the moss knocked off the stone. > > > Don't you also need to implement support for the V4L2 MPEG CIDs ? I would > > expect the driver to support at least the controls used to select the > > MPEG format (MPEG2, TS), even if they're hardcoded to MPEG2-TS. > > That would be possible, for the stream type there is your choice of > MPEG2-TS so that is trivial. There are a very limited set of > standardized controls that can be mapped: wKeyFrameRate, wPFrameRate, > wCompQuality from the VS probe/commit (GOP size, B frames, bitrate). > > Since these controls are optional in the spec, and an overly simplistic > projection of the encoder's actual configuration space, device > manufactures (typically) choose instead to use custom XUs that expose > richer more representative ones. > > Given this state of affairs, I think it would be prudent to blindly > forward the data stream (Which is all, in essence, this patch enables) > leaving the configuration to userspace. > > I'm not suggesting we preclude adding XU -> MPEG2 CID mappings into > uvcvideo later, just that as is this is a valuable step forward.
I agree with this, but I would still implement support for the V4L2_CID_MPEG_STREAM_TYPE control. MPEG applications expect it to be supported. -- Regards, Laurent Pinchart _______________________________________________ Linux-uvc-devel mailing list Linux-uvc-devel@lists.berlios.de https://lists.berlios.de/mailman/listinfo/linux-uvc-devel