Path Space Regularization (aka the solution to SDS paths)

Discussion related to the Engine functionality, implementations and API.
provisory
Posts: 224
Joined: Wed Aug 01, 2018 4:26 pm

Re: Path Space Regularization (aka the solution to SDS paths)

Post by provisory » Wed Sep 04, 2019 12:15 pm

I assume, Dade is perfectly aware of these...

provisory
Posts: 224
Joined: Wed Aug 01, 2018 4:26 pm

Re: Path Space Regularization (aka the solution to SDS paths)

Post by provisory » Wed Sep 04, 2019 12:19 pm

lacilaci wrote:
Wed Sep 04, 2019 11:54 am
What I find interesting is that lower values than 1-1.5 also makes caustics very weak and instead of seeing a lot of fireflies I see just hint of caustics,
...
So maybe caustics are getting sharper, but also weaker and so we cannot really see them?
Lower angle for path regulation = lower chance to hit.

User avatar
Dade
Developer
Developer
Posts: 3138
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Path Space Regularization (aka the solution to SDS paths)

Post by Dade » Wed Sep 04, 2019 1:28 pm

lacilaci wrote:
Wed Sep 04, 2019 9:23 am
some fireflies do appear though...
Are you using GPU rendering ? I haven't yet touched OpenCL code (i.e. you get a mess if you use the GPU at the moment).
Support LuxCoreRender project with salts and bounties

User avatar
Dade
Developer
Developer
Posts: 3138
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Path Space Regularization (aka the solution to SDS paths)

Post by Dade » Wed Sep 04, 2019 1:29 pm

lacilaci wrote:
Wed Sep 04, 2019 11:54 am
I know it's early to some testing but I had a while so I played around a bit...
I have the strong feeling you are using GPU rendering :idea:
Support LuxCoreRender project with salts and bounties

User avatar
lacilaci
Donor
Donor
Posts: 1616
Joined: Fri May 04, 2018 5:16 am

Re: Path Space Regularization (aka the solution to SDS paths)

Post by lacilaci » Wed Sep 04, 2019 1:31 pm

Dade wrote:
Wed Sep 04, 2019 1:28 pm
lacilaci wrote:
Wed Sep 04, 2019 9:23 am
some fireflies do appear though...
Are you using GPU rendering ? I haven't yet touched OpenCL code (i.e. you get a mess if you use GPU at the moment).
Ah yes... I only used opencl.. will switch to cpu.

User avatar
lacilaci
Donor
Donor
Posts: 1616
Joined: Fri May 04, 2018 5:16 am

Re: Path Space Regularization (aka the solution to SDS paths)

Post by lacilaci » Wed Sep 04, 2019 1:40 pm

By the way, will this work with volumes? Cause right now I won't get any caustics in water if it uses homogenous volume (clear works)

User avatar
lacilaci
Donor
Donor
Posts: 1616
Joined: Fri May 04, 2018 5:16 am

Re: Path Space Regularization (aka the solution to SDS paths)

Post by lacilaci » Wed Sep 04, 2019 1:50 pm

Dade wrote:
Wed Sep 04, 2019 1:28 pm
lacilaci wrote:
Wed Sep 04, 2019 9:23 am
some fireflies do appear though...
Are you using GPU rendering ? I haven't yet touched OpenCL code (i.e. you get a mess if you use the GPU at the moment).
Switching to cpu didn't really change anything for me... Pretty much same results as before
It looks like caustics reach a certain brightness and all sds paths keep that brightness/power there are no brigtness or sharpness variances in the pool caustics with this everything is equalized.

no sds:
caustics_simple_190.png
sds 1300 samples, scale 2:
sds_1300.png
scale 0.4:
scalezeropointfour.png

User avatar
lacilaci
Donor
Donor
Posts: 1616
Joined: Fri May 04, 2018 5:16 am

Re: Path Space Regularization (aka the solution to SDS paths)

Post by lacilaci » Wed Sep 04, 2019 2:01 pm

Also, if you look at the pool at an angle, distant caustics don't get resolved at all
It reminds me a lot of the bidir+metro distant darkening problem... related to metropolis?
angledview.png

Fox
Posts: 303
Joined: Sat Mar 31, 2018 11:17 am

Re: Path Space Regularization (aka the solution to SDS paths)

Post by Fox » Wed Sep 04, 2019 5:10 pm

lacilaci wrote:
Wed Sep 04, 2019 2:01 pm
Also, if you look at the pool at an angle, distant caustics don't get resolved at all
It reminds me a lot of the bidir+metro distant darkening problem... related to metropolis?

angledview.png
With the pool test scene around 25m in size, i had far corner shading issue with all the materials and render modes.
Only way to get ride of it was to make bigger than 25m mesh plane (like 100m or so). I made it with invisible null material and moved to the bottom of the scene.

User avatar
lacilaci
Donor
Donor
Posts: 1616
Joined: Fri May 04, 2018 5:16 am

Re: Path Space Regularization (aka the solution to SDS paths)

Post by lacilaci » Wed Sep 04, 2019 5:16 pm

Fox wrote:
Wed Sep 04, 2019 5:10 pm
lacilaci wrote:
Wed Sep 04, 2019 2:01 pm
Also, if you look at the pool at an angle, distant caustics don't get resolved at all
It reminds me a lot of the bidir+metro distant darkening problem... related to metropolis?

angledview.png
With the pool test scene around 25m in size, i had far corner shading issue with all the materials and render modes.
Only way to get ride of it was to make bigger than 25m mesh plane (like 100m or so). I made it with invisible null material and moved to the bottom of the scene.
So you basicaly expanded the bounding box with a fake mesh as a workaround? That's not good...
This is ~42m pool and 1/3 of it is not getting caustics. Before there was ocean mesh 10km wide and this pretty much removed all caustics cause it was too big..

Post Reply