The issue should be fixed now. I have also added in a GML trig precompute step that eliminates up to a couple million trig calculations in the fragment shader, so it should be much more efficient (probably not as efficient as a 2 pass, though). Please tell me if you encounter any more issues!