mesh_resource_marker: re-enabling mesh transparency

[jbohren]

@hersh This highlights the problem you reported in #854 caused by the regression in #752 which intended to fix #751

The issue with the new blending technique is that the base material is always set to be 100% opaque. This patch just initializes the base material based on the marker RGBA values.

[jbohren]

@hersh here's what the mesh marker test looks like with this patch:

[jbohren]

Based on these results, it looks like it still needs some additional logic to ignore the color information. Shouldn't be too hard to fix.

[hersh]

I'm not sure what you mean by "ignore the color information", but the only problem I see is that the alpha=0 rows have a single copy of the original mesh in them.

Thanks for working on this, and including the pictures. I'm really busy right now, but there's always time for looking at a picture. :)

[jbohren]

I'm not sure what you mean by "ignore the color information", but the only problem I see is that the alpha=0 rows have a single copy of the original mesh in them.

Well the tinted meshes appear "washed out" compared to the correct image you posted on #854.

[hersh]

I see, the colorization swamps the native mesh textures, true.

[jbohren]

This is superseded by #893