Laz schrieb: > On Monday 13 November 2006 08:45, Laz wrote: >> On Sunday 12 November 2006 18:54, Martin Wache wrote: >>> Laz schrieb: >>>> I'm glad to report that the newly incorporated "Grab" command works >>>> however, I'm getting incorrect colours. See example grabbed image at: >>>> http://vdr.club-burniston.co.uk/wrong_colours.jpg >>>> >>>> I presume this is another pixel format conversion problem: it's the >>>> same blue cast I was seeing with one of the pixel formats a while >>>> back with cle266 output, although I'm not sure yet whether the >>>> grabbed image comes from the raw DVB stream or after pixel format >>>> conversion for display. >>> Is this using the hardware mpeg2 decoder? It might be that pixel[1] and >>> pixel[2] have to be exchanged in SetupCle266Buffers(). Those values are >>> only used by video filters and grab image when you use the hardware >>> mpeg2 decoder. What happens if you exchange those two and enable the >>> mirror filter? Is the picture still ok? And the grabbed image? >>> >>> The grabbed image is just the last picture which is displayed via >>> DrawVideo_420pl(). It should be before any format conversions. >>> >>> If you don't use the hardware mpeg2 decoder I'm a bit lost... For me it >>> works with Xv-out. >> Aha! Yes, I am using hardware decoding. >> >> I'll have a play tonight to see if I can get correct colours... > > Following up from earlier: swapping round pixel[1] and pixel[2] in > SetupCle266Buffers() gives me correct colours both on the grabbed image and > also on TV-out. N.B. I'm not using any types of image filter, i.e. I didn't > need to mirror. Not sure whether this would make a difference if a > deinterlace filter was used. >
Could you please try if the colors on your TV are still ok when you enable a filter (mirror, or deinterlacer, that doesn't matter) and post a patch if the colors are still ok? If not maybe this is the wrong position to swap pixel[1] and pixel[2], or we have to swap them again somewhere else. Bye, Martin _______________________________________________ Softdevice-devel mailing list [email protected] https://lists.berlios.de/mailman/listinfo/softdevice-devel
