Differences in texture blending with ShaderGenerator vs. fixed-function pipeline about panda3d HOT 3 CLOSED

Thanks for the comprehensive test case! I've fixed this now for the cases you described.

from panda3d.

Thank you for the fixes! It is indeed working quite well now, except for one (hopefully last) remaining difference: any combine mode using CS_constant_color_scale as one of its sources still gives a different result; see e.g. example 4 I gave earlier:

stage 1: M_modulate
stage 2: CM_modulate, CS_constant_color_scale, CO_src_color, CS_previous, CO_src_color

The M_blend_color_scale mode does work correctly now, but perhaps the combine source was overlooked?
Then again, could it be that CS_constant_color_scale is not that well supported by the fixed-function pipeline, while the ShaderGenerator gives the expected results?

Apart from this, do you think it's worth investigating the alpha combine modes as well, or is it safe to assume that a correctly working rgb combine mode guarantees that the corresponding alpha combine mode works correctly also?

from panda3d.

After editing the code sample to take alpha into account, it looks like the FFP ignores alpha when setting the mode of either or both texture stages to M_blend_color_scale; the result looks identical when setting the mode to M_blend (using the FFP).

from panda3d.