Corona Material/Lights Converter (legacy)

[racoonart]

Ok, great! I'll test it out as soon. I also asked ondra if he could provide a better curve to (at least temporarily) convert vray ggx glossiness to Corona - we'll see.

So the tests you made here are Vray Blinn -> Corona GGX or are they Vray GGX -> Corona GGX ?

[peterguthrie]

sorry, missed your post.

The tests I uploaded were Vray Blinn to whatever Corona is

[dubcat]

I made a gradient from 0-1 gloss with 5% difference between each column. The idea was to use it in Glossiness, black diffuse, 100 ior, white dome and compare the reflection pass between V-ray and Corona.
Everything went according to plan, but then I got some weird results with Corona, so I can't compare them. I have no idea if these graphs/curves are useful, but I thought I could share them in case they are.
You can see that pure black is around 0.15y V-ray GGX and 0.24y V-ray Blinn. 0.24y Blinn is the same as around 0.29x GGX on the graph. We can look at the original gradient and find the column  that is 0.29x on the graph. This is the glossiness value of GGX that is 0 Blinn. If only the Corona test worked.

This is the glossiness texture I made




V-Ray GGX with default 2.0 tail




V-Ray Blinn




V-Ray Ward




V-Ray Phong




This is the weird result I got with Corona. At first I thought it was pure white.



But then I applied a Gamma Correction and got this result.
Looks like the walking pattern of a drunk.

[racoonart]

When I read your post yesterday I wasn't sure what you did and what those graphs represented (I somehow didn't realize the dome light part in your description). I then had the "original" idea of measuring the luminance of reflections to make a mapping between the BRDFs. So, you probably planted the idea into my head without me noticing :D Thanks!

So I made a 100% reflective plane, a tiny light emitting sphere above and a camera on top - all 3 in a line, the camera facing directly onto the plane (to have no angles, different renderers seem to have different approaches for pixels at sharp angles).
I then wrote scripts to measure the pixel luminance right in the middle of the reflection for each glossiness step and made tables of glossiness <-> luminance for vray ggx, vray blinn and corona ggx. It went all well, the curves looked fine and I was able to build conversion tables between the BRDFs.
Below you can find images with VrayBlinn on the left (1.0 to 0.55 glossiness in 0.05 steps) and the CoronaGGX result on the right using the conversion table.

I'm also a bit surprised that it works so well...

[peterguthrie]

nice! so will this find it's way into the next version?

[racoonart]

I hope so. Next I'll have to find a way to make curves out of the tables so I can also support glossiness maps - this may be a bit more tricky.

[peterguthrie]

ah yes... maps.

Rather you than me!

(tell me where the paypal donate button is btw. its an already fantastic script)

[racoonart]

ah yes... maps.

Rather you than me!

I'm currently doing it all by hand. First plotting the table, then I do a custom curve control script where I can type in and try out the positions and in/out tangents of the points. After each change I have to make a screenshot and do a comp with the plot in the background and the curve control as a layer on top and see where I have to improve it. See the attached image below.

(tell me where the paypal donate button is btw. its an already fantastic script)

Oh, thanks a lot :) I didn't bother to include a donate button anywhere on the website because I didn't work in the past. If you want to donate something, I have one on my ExtendBorders script download page ( http://www.racoon-artworks.de/?p=343 ) . Unfortunately, to this date, even though it has 6k downloads, the script made less then 10$ (2 people) in donations ;)

[racoonart]

Ok, glossiness map conversion seems to work fine. Attached image: left is VrayBlinn, right CoronaGGX
So, If someone has a representative production scene that uses lots of glossy materials and wants to try out the new conversion method, drop me a line via PM and I'll send you a version. Every piece of feedback about this is important! I want to make sure the results are good before I release it :)

[maru]

I notified the Evermotion guys, they were interested in glossy conversion improvements. Let's see how this evolves.

[dubcat]

You're the man Clown!

[racoonart]

New VrayLight Directionality conversion using the same luminance measurement method
Directionality 0.3 -> 1.0 in 0.1 steps

[antanas]

Wow ! sadly I've got no recent\decent vray scene to test the new conversion on but still wow.

[dubcat]

Here are some test results with "v1.19 beta 02".

3 Evermotion scenes. I have only used the converter and hit render, nothing is modified.
Each scene has been rendering for about 30min on an i7-4770K.





Shaderball tests.




Here is a graphical representation of what the converter is doing.



Looking good. I think the big difference in the kitchen render is due to the mirror.

[arrival]

where i can find the v1.19 beta 02 version.?  i see only 1.18