Does this help? If you're going to do all of those Multiply & Adds, use a ChannelBlend (Summed) instead. Also, consider what your math looks like outside of the 0-1 UV range. For instance, if you're wanting each quadrant to repeat past the 0-1 range, you're going to have to do a lot more math than if you just want something like a checkerboard repeat. And if you're going outside of the 0-1 range anyway, you might just want to work with positive/negative Us/Vs. +u+v is one texture, +u-v is another, etc. I might be really misunderstanding what you're going for though. I went to answer this, but when I checked my assumptions, I was wrong. I'd suggest trying out each by just looking at the stats section in the compiled shader source and testing it out in a stress-test scene. Depending on the platform, Lerp (Mix for glsl) can yield fewer instruction counts than summed, but that doesn't mean it'll be faster. Oh, and using the channel blend with a layered setting is the same as using lerps. [EDIT::] Consolidated several posts into this one.