LuxCoreRender Forums

You remenber My last renderer war Cycles vs luxcore. Just now wanted to make a Luxcore vs Octane now but that would be really really destroying and pretty unfair to luxcore. The reason behind : clamping

While Luxcore and cycles tend to loose lighting quality and information Octane is able to keep almost everything even in very low clamping value scenario.

Octane Clamping : 1000
Octane Clamping : 100
Octane Clamping : 10
Octane Clamping : 1
As you can see at same render time (5mn) you get a clean image and this is without any caching. In such extreme clamping case Luxcore would loose lot of indirect and direct light information as clampling is cut the same way for both.

Maybe some fireflies filtering like in clarissefx can help.

https://www.clarissewiki.com/4.0/firefl ... ering.html

Can you post noise-free images without clamping vs. clamping = 1?

It is likely to be a clamping method based on the PDF of the path, not on a window around the expected value.

B.Y.O.B. wrote: ↑Mon Mar 23, 2020 4:05 pm Can you post noise-free images without clamping vs. clamping = 1?

40Mn Render no clamping aka (1000000)/ it is the limit

Dade wrote: ↑Mon Mar 23, 2020 4:20 pm It is likely to be a clamping method based on the PDF of the path, not on a window around the expected value.

Clarisse Fx or Octane one ? is it better If yes can we add it ?

Sharlybg wrote: ↑Mon Mar 23, 2020 4:56 pm

B.Y.O.B. wrote: ↑Mon Mar 23, 2020 4:05 pm Can you post noise-free images without clamping vs. clamping = 1?

40Mn Render no clamping aka (1000000)/ it is the limit

Not noise-free, but I guess it's enough to estimate the final brightness.
Clamping = 1 loses quite a bit of brightness compared to this, from what i can see.
Would the clamping in LuxCore right now be much different?

Clamping = 1 loses quite a bit of brightness compared to this, from what i can see.
Would the clamping in LuxCore right now be much different?

Of course there is a small loose but compared to the speed and preserved light quality it by far Better.

Look luxcore at clamping 1 took 12 minute and stay noisier:
Clamping at 0.2 is less noisier and took 7mn(compared to 5mn clean octane) but lighting is crashed :
Seriously they don't handle clamping the same way and i can feel it quickly while working with it.

Edit i messed up resolution.

but you can see how it look.

Note there is no Caching on luxcore of course.

Luxcore + caching clamping at 1 is still the boss but it isn't ideal to be match in speed when you run Gi caching by a Brute force solution.And Octane isn't using optix in my test

I think between Clamping MIS and other method there is lot of room for improvement.

Brute force unclamped 50 mn luxcore :

Sharlybg wrote: ↑Mon Mar 23, 2020 4:56 pm Clarisse Fx or Octane one ? is it better If yes can we add it ?

If I add the their sources...

The PDF based clamping is an old JeanPhi's idea, he tried an implementation based on local vertex path clamping and it didn't worked well however I think the idea may work if considering the complete path PDF.