Quoting Ville Syrj�l� ([EMAIL PROTECTED]): > On Thu, Mar 20, 2003 at 05:32:04PM +0100, Denis Oliver Kropp wrote: > > Quoting Ville Syrj�l� ([EMAIL PROTECTED]): > > > For palette lookups in the accel code I've used 0x80 as the alpha value > > > since palette alpha is irrelevant for this format. > > > > We should add a palette capability if the palette uses the alpha component > > and have a seperate lookup method for palettes without alpha. > > Good idea. It would also improve the accuracy of the palette lookup since > the alpha component wouldn't affect the calculations. > > But then there's a problem with SetColorIndex() since it can't change the > alpha value. Add a new SetAlpha() method or instruct users to use > SetColor()?
One would have to use SetColorIndex() and SetAlpha() then. Sounds good. Or use SetColor() which does a lookup for rgb and sets alpha directly. > Also I was thinking about using the alpha from the first two palette > entries for c2bpp15lalpha and c2bpp15halpha. These would allow more than > on-off opacity in ARGB1555 mode but only two levels at the same time. I > haven't really tested the sub-picture stuff with CRTC2 in other than > YV12/I420 modes but I think it's only possible to select a color and > opacity. So there's no sub-picture surface only a constant color. There's > a table of acceptable settings in the G400 specs. I'm not sure if anyone > would seriosly want to use this feature. The sub picture layer supports ALUTRGB1555? 1bit alpha lookup plus rgb15? If it simultaneously supports color keying, too, it would exactly meet the minimum platform requirements of MHP ;) Because it requires per pixel alpha with at least three different transparencies (0%, 30% and 100%). -- Best regards, Denis Oliver Kropp .------------------------------------------. | DirectFB - Hardware accelerated graphics | | http://www.directfb.org/ | "------------------------------------------" Convergence GmbH -- Info: To unsubscribe send a mail to [EMAIL PROTECTED] with "unsubscribe directfb-dev" as subject.
