I created a square plane. Its lightmap UV is just a box spanning the UV space, snapped to the edges. Due to the sheer simplicity of the mesh, I thought that using a lightmap resolution of 4 would be perfectly fine, but when I baked the lighting the results were horrendous; two identical planes put beside one another looked like they had two different colours. This was present even in production quality lighting.
Thus I cranked up the resolution until I reached 32, which gave a perfectly smooth output in just preview lighting–but only when the planes were ‘outdoors’ (indoors being them facing inwards like in a room), whereas indoors planes required a resolution of 64. The part I don’t understand is why this matters. There is literally nothing going on in the mesh so it’s just one UV island–perfectly square for the LM, and yet I get bad results with lower resolutions. Does those planes really need 64 pixels? I’ve always thought that the resolution should be whatever is lowest that can fit in all the islands with padding and etc, but am I misunderstanding lightmaps in that I should be giving large areas a larger lightmap, no matter how simple they are?
Shadow Seams/Shading Differences with indirect lighting:
“I’ve always thought that the resolution should be whatever is lowest that can fit in all the islands with padding and etc, but am I misunderstanding lightmaps in that I should be giving large areas a larger lightmap, no matter how simple they are?”
Lightmap resolution is dependent on the efficient use of the 0,1 UV space when packing the lightmap UV along with the size of the mesh in the scene. Having a large mesh with a low lightmap resolution will generate low quality shadowing. The idea is to keep the lightmap resolution relative to other surfaces to make sure to keep it low enough for your needs and quality bar so that the lightmap density is consistent for the scene. Otherwise, it can break the consistency of the bake. Use the Lightmap Density view mode to check this.