I’m trying to achieve near “photorealistic” rendering with full shadows. While spot lights work pretty fine with self-shadowing (see the image below), point light shadows looks not as good - it produces totally shaded objects (next one). Demo: https://playground.babylonjs.com/#U2F7P9#4
Apparently, the omni-directional cubemap is used to produce shadows for point lights. I can substitute point light with eight spot lights (for testing only, 'cause I cant introduce additional seven lights) and receive suitable self-shadowing.
I assume there should not be difference in these images, because second example is actually one side of the cube map, which is lit by one ray of light, which is emmited from the same position. Generally speaking, I expect same shadows on both images. But they’re different. My question is - Why?
Or, may be different approach exists to get full shadows on the scene?