in general if you look at most algorithms that get widespread adoption in graphics it is because they cover the highest % of use cases. for instance you can do depth sorting in a variety of ways. but zbuffer is used 99.9% of the time because it is simple, efficient and works 100% of the time. its sort of a Darwinian…
so to come back on a couple of the points made first is regarding having to get the full res LOD for the calculation, unless the silhouette is completely different from LOD - LOD then it wouldn't matter, if the lower LOD starts taking up too much for that res, then step it up to the next LOD. the second, regarding pushing…
Really interesting Idea, but the problem would be that you always have to render the complete scene (without lodding?) for every frame and do a really heavy-weight pixel-calculation stuff. checking how many pixels on the screen have the same colour will impact your performance (your worst case would be cheking all pixels…