there's something odd with the way the vex layered surface shader works with images that have alpha's..

it only seems to be making a layer correctly masked if the rgb has matching black areas as the alpha. i don't know why this is but seems pretty dumb. it shouldn't matter what my rgb is doing and should just be looking at the alpha itself to mask upon. i've tried numerous formats (tga, tiff, rat) with all the same results.

i also tried it in vops and there is no problem there. Vops didn't seem to care what was on the rgb and just used the alpha as it should..

why does the layered surface shader do this? any thoughts or solutions?

thanks,
dave

also seems to be an issue with the vex supermat shader… is acting the same exact way as the layered surface.

it looks like they are both seeing the alpha, but seems to be adding or doing something else with it to the underlying diffuse color as opposed to comping it over. (and i have it set to image over). seems to work fine with the images that have matching black rgb and alpha though.

Those shaders probably assume that the image was properly pre-multiplied. This means that no R, G, or B level can be higher than the A level.

That is fairly standard and necessary for most compositing software too.

If you created your RGB and your A seperately, just multiply the RGB times the A and put it back into your RGB channels. Or create your own VOP and do this.

-Craig

ahh.. that makes sense craig.. thanks.. i guess im used to having the choice as to interpret the alpha as premultiplied or straight (unmatted) in compositing apps…

i'm guessing that these shaders don't have interpreting options.. atleast i can't find any..

thanks,
dave