Env. Light Visibility Cache

[kintuX]

Well, that's the Conservation of energy. And it's simply in the nature of existence. In law of economics, physics, rendering, cooking... you name it. There's simply no way to have both in same amount of time. It's either quality or quantity.

All of which have been explained and shown many times... last time here, roughly a month ago.

Why do we always need to get dumbed down and biased? Why don't you simply use biased engines then?
Another thought to accompany your dreams and goals in life: Con-Fusion brings Distraction brings NOISE.

[Sharlybg]

We are happy with what we have with Luxcore it is GREAT and FAAARR from a underperforming. But it is in computer graphic tech to push boundary and smah ceilling to create performance and Unlock creativity.

This is the kind of behaviour that lead to the Luxcore of today. If no one mention it and complain about something thing stay like they are.
Before Lacilaci start to point out how bad direct light performance is compared to cycles i had never imagine something like Envcache performance jump can be possible again after PGI.

Now i can hardly accept to work without Envcache speedup. This mean that in case i work on project with highly reflective large surface ( floor/ceilling/wall) my performance will go down to stone age rendering. If there is any genius idea to break that i want it. if there is no way to escape i will still love my FAVORITE ENGINE and try old school cheat to speed up again .

# NEURALINK MOD OFF

[Dade]

You are missing the point: you don't do direct light sampling on a perfect specular surface (like a mirror) because it is useless, the proposed ray direction will always evaluate to black by the specular material (i.e. it will never match exactly the only direction a mirror can reflect).

The glossiness threshold just make glossy(-nearly-specular) material to work like a perfectly specular material and avoids to do direct light sampling (it is useless anyway).

It is not a problem related to portals, env. cache, etc. It is related to glossy materials. For the same reason, you can hardly see the highlight (i.e. the white dot) of a point/sphere light on a glossy material because more you reduce the roughness, more direct light contribute is nearly zero.

Just try it: a sphere light source (point light source with a radius) and glossy sphere, more you reduce the roughness (i.e. more nearly specular it is) and more the glossy sphere will be black (with a lot of noise where the highlight was).

[kintuX]

Also,

This engine is already performing above expectations. Look around and try others - you'll be up for a waiting game.
Some conformation has to be made in order to move on. Otherwise development can get stuck on such minute issues.

Anyone should work with a clear mind, ready to prevent and avoid such things while working. As we get to know this issues and learn, don't you think it's stupid, a farce or even evil to intentionally repeat same actions? Doing same mistakes over and over?
While we could do other, better things. Like you do videos.

So, would you care to (in another thread) make some short and simple, project based show-offs like documentaries (not tutorials). Showing and explaining what to take care of. Some tips & tricks like ones mentioned here. There are so many things in Lux that work very well. That's what we should worry about. To get viewers attracted and hooked to Lux.

[lacilaci]

Why the extreme noise then? And what can we do about it.

Is it glossy shader itself? Or is it a user error that can be avoided? Maybe a convenient workaround?

[Sharlybg]

Why the extreme noise then? And what can we do about it.

Is it glossy shader itself? Or is it a user error that can be avoided? Maybe a convenient workaround?

About that. With your experience with CORONA how it handle such scenario cases ?

[Dade]

About that. With your experience with CORONA how it handle such scenario cases ?

As afar as I know, it doesn't handle the situation in a different way ... as far as I know, there isn't another way to handle it. Try to do some test but like I said, this is not a portal/env. light cache problem, it is a glossy related problem (for glossy materials with very low roughness).

[kintuX]

Yup!
In dark places, with low light, lots of glossy surfaces it also takes looong to clear. So you're better using 0 value rather than 0.0123.
Also, cameras and even human eyes have hard times gathering light. Tho our experience & intelligence fills the gap. And sometimes, imagination takes over the truth and reality - mind creates hallucinations. Just like there's no pure white. We made it up

[lacilaci]

Hm, I can't test against corona right now. But I don't remember having problems with glossy objects at all.

As far as other renderers go, the closest I can get to comparison with this scene is this with cycles:

it's ~6 min. rendering, luxcore was 8 but I cannot set precise time in cycles. Close enough.

Scene is also a bit different with lighting etc, but the problematic area (sink) is pretty much just as occluded and material is also pretty much the same when it comes to parameters. As you can see there is not much more noise than other areas in the rendering (and this is without some GI caching or any cycles's AO fake settings)

Maybe it's some biased performance tricks or something that others do to avoid the issue or I really don't know.

here it is as metal at 0.01 roughness with the same sample amount as before (128)

this is pretty much my experience with other software too, maybe back in the days where in vray or mental ray I had to set manualy samples per object I would run into some objects being much noisier. That's looong ago...

If it's not something fishy on luxcore's part, then they must use some voodoo tricks.

I'll try to play with the shader or some settings in luxcore cause even in luxcore I don't often run into this... But when I do I have to switch over to cycles cause that noise is not clearing up!

[epilectrolytics]

Maybe it's some biased performance tricks or something that others do to avoid the issue or I really don't know.

Maybe in Cycles filter glossy helps with that.
Do you have reflective caustics and filter glossy activated when you render in cycles?

I never found an explanation how it actually works but I suspect it uses higher roughness for second and further bounces or something.