Yeah, I was wondering about that (and thinking of trying that next). I think ambient occlusion, although it comes at a bigger cost, adds more realism on a bigger scale than contact shadows. I presume if you have two neighboring hillsides or canyon walls, one hillside or wall is in shadow but the opposite hillside or canyon wall is in brilliant sunlight, the shaded hillside and anything down in the space between the two hillsides/canyon walls is going to be lit/shaded in ways that are not accounted for by contact shadows - the Avri Lab blog post that I cite gives the example of objects that are out of the field of view affecting ambient lighting. I’ve noticed the sort of effect that I’m talking about in flying around Mann Gulch and Machu Picchu, both places have steep, rugged terrain with many nearby surfaces reflecting light and casting shadows but nothing really in contact - it’s all light and shadows dispersed over 100’s or 1,000’s of feet of distance onto other surfaces. Perhaps for trees on hillsides, yes, I can see contact shadows might enhance visual perception in relation to the connected ground but I think in a stand of trees or even for leaves on different branches of a tree, you’re going to have a light/shadow effect radiating through space that’s not a direct projection of a contact - aren’t contact shadows a much more narrowly locally defined effect? - but it’s my vague newbie understanding of ambient occlusion, too, that AO is most important in perception in close, confined spaces, e.g., a hollow tube in which shadows should get darker the deeper one peers into the tube, etc. Anyway, thanks for your expertise. I’d be very interested to learn the difference between contact shadows and ambient occlusion. As azmidi’s VR Bang-For-The-Buck analysis shows, implementing contact shadows costs far less than AO (an observation also consistent with the more limited scope of Contact Shadows dealing with the realities of visual space).
Row copied from azmidi’s Google Sheet link in his posted analysis:
Edit_Update: What I’ll do when I have time, too, is I’ll generate an identical image to the two that I’ve posted in the Everything Looks More Real With Ambient Occlusion in VR But Performance Takes a Hit, Obviously thread but make the new third image one with no AO but Contact Shadows instead. I’ll add that image to the group posted so that one can click and cycle between large versions of ~the same image to see comparatively what kind of difference using Contact Shadows vs. AO vs. neither makes to the scene.