Tronic
15th July 2004, 05:11
The NVIDIA video overlay (and probably others as well) seems to only use nearest neighbor scaling for downscaling the 1920x1080 video to a smaller video window. This produces bad quality and unwanted artifacts. Is there any way to get better scaling, but to still preserve hardware YUV-to-RGB conversion?
The filter should take weighted average of all the source pixels (aka texels) that fall under the pixel on screen. The texel weights should be defined by the portion of the screen pixel that the texel fills. All this should be done in gamma 1.0 (so you need gamma conversion before and after).
Is there even software that could do this (and fast enough for HDTV)?
What is this algorithm called anyway?
The filter should take weighted average of all the source pixels (aka texels) that fall under the pixel on screen. The texel weights should be defined by the portion of the screen pixel that the texel fills. All this should be done in gamma 1.0 (so you need gamma conversion before and after).
Is there even software that could do this (and fast enough for HDTV)?
What is this algorithm called anyway?