Chaos Corona for 3ds Max > [Max] General Discussion

How to handle backplates?

<< < (2/2)

dj_buckley:
Another issue is that backplates in environment overrides, don't play nicely with ACES OT or LUTS etc, and the tone mapping control rarely works to reverse the effects.  So if using any tone mapping techniques in the frame buffer, then its pretty pointless getting your backplate looking how you want it to in your photo editing app beforehand, as it rarely looks like the same in render.

If you process backplates as you would a normal photograph and then use it in Corona with any sort of of Lut, ACES OT or tonemapping tweaks, you'll effectively be applying two tone curves to the backplate (one in Lightroom, one in corona) and one to the render.  Giving a mismatch.

I'm currently experimenting with processing backplates using Linear Profiles and using them that way, almost akin to prepping a proper Albedo texture.  That way the only tone curve being applied is in render and theoretically the backplate should align with the rest of the render in terms of 'look'.

Quite often we capture full 360's on site to use as reflection overrides and/or direct visibility overrides - but as already stated, Corona in it's current state doesn't really cater well for any workflow involving jpeg backplates in render.

maru:

--- Quote from: aaouviz on 2021-08-18, 14:42:29 ---Problems with option A:
If I want this city skyline visible through glass I must also set the 'refraction' (and probably 'reflection') overrides. This is problematic as it creates the noisy bright spots all over the image. Not ideal. I know corona devs are aware of this problem, but I'm pretty sure it's not fixed yet, correct?

--- End quote ---

This should have been fixed by:

--- Quote ---Fixed issues with environment overrides, such as increased amount of noise and incorrect results

--- End quote ---
in Corona 8:
https://forum.corona-renderer.com/index.php?topic=33839.msg190180#msg190180

If you are still getting some similar issues with using this approach, could you please share a scene with us through this link? https://support.chaos.com/hc/en-us/requests/new


Other than that, I agree that rendering background plates is difficult now. Especially after implementing the Corona Physical Mtl, which always has some glossy reflections, so the background is always reflected at least a bit.
I think the "safest" solution is using environment overrides in the Scene tab. Using a Light Mtl with "emit light" off will most likely give you the "dark highlights" issue described here - https://support.chaos.com/hc/en-us/articles/4528311915025-Known-bugs-and-limitations
It is similar with using a self-illuminated material with rayswitchers. Generally, this is a kind of fake that Corona handles in a way that is expected from the render engine / light simulation point of view, but not for the users. We have this properly logged with example scenes and it is acknowledged, but the main issue is that resolving it without breaking other stuff will be difficult and will take a lot of time. Hopefully it is another case of something nearly impossible that will be done one day.

Other than that, *sometimes* some workarounds are possible, for example using a combination of "emit off" and "emit on" with a rayswitcher: https://forum.corona-renderer.com/index.php?topic=32357.msg181607#msg181607

vemod:
I usually create an alpha for the sky and cut it out from any backplate, and then I do a modified texture that looks like a proper albedo texture for the stuff that's left.
If I feel real fancy I will extract reflecting parts (say windows, etc) and do a separate texture of that and place it in metalness/reflection channel.
I then assemble it either on projected geo (the best) or just a simple plane.

It won't look identical, and usually requires some back and forth until it looks right, but it responds much more natural to any tone-mapping shenanigans.

Removing all lighting info is tedious at best, I have had great success with 3dLutCreator and the volume channel for quick fixes though.

Navigation

[0] Message Index

[*] Previous page