In my test case, the quad isn't actually overlaying anything -- it's the only thing being displayed on the screen (aside from pictures of a young and vibrant Bea Arthur).

However (and my knowledge of shaders is limited), wouldn't that bit of shader code be called the same number of times even if it were a smaller bitmap stretched to 1024x768?

Humm. But in this case (view entity with .material set), isn't the shader applied to each pixel as it's drawn to the screen, rather than to the camera's view when it's drawn to the view entity's bitmap?

Ah yes, you're right. So that won't have much of an effect on performance.. I guess that if you would use a smaller view entity (as in size on the screen), and then render that view entity to a bitmap, and display that on the screen in a scaled panel or something, you could get some performance gain. However, the overhead of rendering again might diminish the performance gain. On top of that, it's currently very difficult to use the result of one postprocessing shader in the next one properly.

I'll see if I can give that a try; I've always wanted to figure out how to copy and scale bitmaps quickly.