In this project, I’m taking a fully dynamic lighting approach; with no baked lighting at all. I’m using CSM+DFS and DFAO+SSAO. The directional light, with the csm and dfs, is set to stationary. Does it make a huge difference if it’s set to movable? Also, the skylight is set to movable and has a non-zero occlusion max distance set(300 in my case).
The CSM and DFS are handling the actual “shadow” shadows of the scene. My CSM is set with a distance of 2k and the DFS is set for 10k. So that means between 0-2k, I use the CSM and between 2k and 10k, I use the DFS(which uses the DFAO mesh data). Beyond 10k, the DFSs disappear.
Now for the DFAO/SSAO, I’m using the DFAO to give all of the detail shadowing, from a distance, and the SSAO for all of closeup AO detail. On the skylight, I have the occlusion max distance set to 300 because I don’t want a whole ton of performance eaten up by unnecessarily long tracing. The SSAO is set to have a fairly tight radius (20-30), for “filling in the cracks.” The SSAO fade out distance is set to 1500 and the fade out radius is set to 400(smooths the fading); so that it doesn’t try to compete/overlap with DFAO.
So please correct me if I’m wrong in using this workflow. It seems to produce some pretty amazing results. The DFAO can even be turned off, to save performance, and the scene still looks pretty decent with just CSM+DFS+SSAO.