High I’m doing some tests with the reflection probes. I notice you can set the reflection bounces up to 5.
Is it possible just to have infinite reflections? i.e. It could just use what was previously calculated for all the other shiny surfaces when calculating it’s new reflections? Do you see what I mean?
So after frame 1 it would just be like having one bounce. Then after frame 2 it would be like having two bounces and so on. It wouldn’t take any extra processing power since it only calculates one bounce each frame.
If you think about what’s involved… one ball looking at one other ball, that would be one probe… but to get the reflections that are ON that other ball, you have to look up that other ball’s reflection probe also … and may see ball 1 reflected in it. Okay so far… but it now depends on which order the objects are drawn… ball 2 only reflects what ball 1 looks like BEFORE ball 1 gets updated with the reflections that it found from ball 2… so now there is a conundrum. Even with that said, each time you want to get an extra bounce, you have to look up another reflection probe (I’m guessing) and every time you do this it all gets slower performance-wise (since they are texture lookups). What you’re asking for is to have like 50 bounces of reflections so you see like 50 little counter-reflected spheres within spheres using some kind of feedback mechanism… like looking in a mirror reflected in a mirror… but a) this would still produce a time delay between frames/bounces (you add one new bounce per frame, takes 50 frames to update the entire sequence) … and there is not unlimited processing power. You can’t keep results of previous bounces UNLESS all objects are absolutely stationary, otherwise you have to recalculate eveyy pixel in every bounce in every object. Even in real raytracing this is a similar problem.
Nah. If you point a mirror at another mirror, there is a time delay caused by the speed of light. Or if you point a TV camera at a television there is a time delay. That is fine. The inner reflections will only be wrong by 1 frame. And the inner-inner reflections by 2 frames etc. It would create some interesting feedback.
with reflection probes you cannot achieve infinite reflections. It may be possible to do this with realtime probes, but not with Baked or Custom probes.