Gerd Knorr (kraxel@xxxxxxxxxxx): > > Interesting. Should I call VIDIOCSWIN before VIDIOCGMBUF ? > > Better use the v4l2 API, this hasn't those hidden features :) I guess that's fair. :) > In theory VIDIOCSWIN / VIDIOCSPARM (palette field) should be used to > set the capture size format. It is required for capture via read(). > Drivers might use it to calculate the buffer size for VIDIOCGMBUF. In > practice this is nearly nowhere used that way because bttv doesn't > need this for historical reasons. Requiring applications to do that > now would likely break lots of them. Sure but I'm willing to do anything to ensure that I work with the largest amount of cards with minimal pain. I guess there likely aren't any other drivers that do it like this though, right? > > > I've changed the default capture size in the latest snapshot. > > > > To the maximum capture size? > > 768x576 @ RGB24 Sounds good. -- Billy Biggs vektor@xxxxxxxxxxxx