Excluding Lighting objects origin from the rest

Started by Kadri, March 21, 2021, 10:08:50 pm

Previous topic - Next topic

Kadri

I think a setting in the light nodes to exclude specific objects, clouds, atmosphere could help here.
In my case here probably excluding atmosphere.

Dune

Yes, like in the sun. That would be for Matt to implement.

Kadri

FRUSTRATING!

With card object luminosity test:

Luminosity : 1003101  Render is all white.
Luminosity : 1003100  Render is normal but no lighting.

Small crop and big crop in the viewer looks different too. And even the opposite what you get in the final render...

The luminosity changes i think depending on the object size and you begin to play ones again with settings.
I think i got light with much lower settings first but got to those high numbers while playing around with this strange appearances.

I give up. I will just try to get rid of those parts in post and try to cheat and hope it works.
I think this will be even easier then playing cat and mouse with all those settings.

Kadri

Just saw that i could get even 3 different looks (full white, partial lit and without light) by just trying to render 5-10 times the same frame...hmm

Kadri

March 22, 2021, 10:53:25 am #19 Last Edit: March 22, 2021, 11:20:08 am by Kadri

I used a lower luminosity like 10 000 and got 3-4 different looks every time i rendered the same frame.
Luminosity isn't reliable. Don't know if it is because of these high numbers or the way it is sampled.
I have a feeling that AA, detail and-or GI settings will affect this too...
So until i know what i really should do i won't use this method but it was worth a try.

Dune


WAS

May have to play with lighting in atmosphere, and lighting on surfaces with the large sphere/cube.

Kadri

I thought that maybe the card object is a big flat polygon maybe the luminosity was not accurately sampled or so.
So i tried with a high poly obj object. But couldn't get it work like i wanted. Testing in the real scene takes too long.
So i will wait and see if you or others will get any good results.

But still curious. I will see what happens in a default scene.

Kadri

This test is with an imported kinda high poly flat sphere.
Luminosity is at 100. It works with 1 too but it is seen with 100 better.

This is an animation of 25 frames (1.6 Mb gif). I didn't animate and didn't change anything from the defaults.
So this is actually the same frame rendered 25 times.

Kadri

March 22, 2021, 09:45:42 pm #24 Last Edit: March 22, 2021, 09:55:53 pm by Kadri
I tried many things, GI, more atmo samples, AA, animation settings, subdivide etc.
With some it looks less sporadic. But still it is like the sampling is all over the place.

If nothing comes to my mind this is so far i go. The object polygons could be a problem (doubt it but just in case) for example.
I will triple them and try ones again.
Edit: No. Same results.

If you want you can try the same scene here too:

WAS

Have you tried using a saved cache of one frame for all frames? I think this is random "noise" introduced by the approximation of the luminosity lighting and cannot be helped. It can get a little better forcing it to sample better but will never be accurate and takes some shortcuts by way of approximation.

This is why I don't like luminosity and how it works. Very fake. Like with my red dwarf, I got to a point where it looked alright, but if you look closely the rings have the same circle effects where it was approximating lighting coming from the clouds luminosity and underlying planet luminosity.

Kadri

Quote from: WAS on March 22, 2021, 09:49:47 pmHave you tried using a saved cache of one frame for all frames? I think this is random "noise" introduced by the approximation of the luminosity lighting and cannot be helped. It can get a little better forcing it to sample better but will never be accurate and takes some shortcuts by way of approximation.

No. I haven't tried that but that could work. But wouldn't help in a real scene unfortunately (?).

Kadri

Jordan i just tested caching like you said. It works. The lighting is blotchy but not going all over the place.

The sampling is like the subdiv settings different and lower then the rest probably.

I think here is the place where Matt says something ;)

WAS

It would be cool to see improvement here, at least maybe some sort of slider for accuracy, despite hitting performance, to leave to the user to find a sweet spot.

Kadri

Yeah. Maybe Matt didn't see it coming that luminosity would be used in these kind of kinda extreme ways...
Hope he changes this like you said (if there isn't already a way of course).
It could be kind of an alternate use instead of lights in some situations.