Fresnel reflection to dark?

[Micha_cg]

Fresnel reflection to dark?

Here I'm again with my fresnel problem. Today I have found an interesting effect.

But first an example photo of the effect I like to get. A black glass sphere in an outdoor environment - the heaven is good to see everywhere on the sphere.
image1


Here my new test rendering: left side standard glass with fresnel reflection. For me, the reflections are looking much to dark. The right side show the same material, but the IOR of the refract color of the fresnel reflection control is now 1. Looks more like my reference photo or?
image2


Could it be, that the glass reflections are so dark, because some options of VfR are internal wrong linked? Maybe a bug of VfR and not of Vray?

Here an example with visible refraction - the refraction IOR of the fresnel reflection is set at 1 and the IOR of the refraction layer is 1.55.
image3


Now I'm very curious.

Greetings
Micha

[Micha_cg]

Fresnel reflection to dark?

Seems to be a good day to hunt bugs.

Here my test scene again. I have delete the refraction layer and set a diffuse layer. Reflection is set to fresnel 1.55/1.55. Left side black diffuse color and right side grey diffuse color. This is an other old problem - dark diffuse colors kill the fresnel reflection, but it should be independent.

image4


Next bug: if I render the whole scene the render time is 40s. If I render the window, so that the grey sphere is visible at the right side, than the render time is 1min32s. Render window slow down the render speed? If I render a window, than merging the light cache passes and prefilter light cache need extrem much time. This must be a bug, because if I would scale down the rhino viewport and zoom to the view of my "render window", than I suppose so the render time would be 30s only.

Back to the fresnel bug. I have found a new effect. If I render a grey sphere with fresnel 1.55/1.55 and with fresnel 1.55/1, than both images are indentical good looking (like the grey half sphere in the attched image). But if I render a black sphere with 1.55/1, than no reflections are visible, the sphere looks like a black emitter.

[Micha_cg]

Fresnel reflection to dark?

... and to day I'm working on a scene with cars in front of a hotel. All dark colored cars are looking like matte painted. Please try to fix this bug so fast as possible.

[joe_b]

Fresnel reflection to dark?

When you put the fresnel map on the reflection map it is like transparency. So if you have a dark diffuse color - it will be dark.

An IOR of 1 results in complete transparency - there is no fresnel effect at all in this situation, if I recall correctly. So you are just getting a straight up reflection all over. I'll verify this on Monday.

When I get into the office Monday I play around with this stuff some more. I think with the changes to how the multiplier works you should be able to get results that you want.

Perhaps i can look into making my own Fresnel texture to see if the results come out the same as the current one. I don't know when I'll have time to do that though.

[Micha_cg]

Fresnel reflection to dark?

When you put the fresnel map on the reflection map it is like transparency. So if you have a dark diffuse color - it will be dark. ...

I don't understand you, but I know that black painted cars has reflections. I'm sure.
... and my CG cars are so bad looking.

Look at my last posted image (grey/black plastic) - this is not natural. Other example: do you know this wonderful black concert pianos? Totaly black but nice reflections. The reflections of the finish are independent from the color below!

[Micha_cg]

Fresnel reflection to dark?



OK, I have done some more research and find a solution - try to be physical correct!!! Two things are important:
(1) use a low burn value so that the color mapping works approx. like the human eye, that can see high contrast without problems.
(2) use physical correct colors: natural white has a reflectance of max. 80%
(3) the same for black color values: best results for black with intensity 20..10

So I have set the burn value at 0.8 and set the white to 190. This allow me to increase the environment intesity from 2 to 12. And so the cars looks much more like I have expected .

Time for a new tutorial or?

image5


PS: But a reflection multiplier would be good anyway, good for the CG freedom. Interesting: the problem of image4 is solved by the chnage of the color mapping burn value too!

[joe_b]

Fresnel reflection to dark?

I'll give you a new TexFresnel that uses multiplier properly. I'm guessing that will probably solve your problem (or at least provide a way to get better reflectivity ). If that still isn't looking correct then I will make my own TexFresnel from one of my books and we can see if that behaves in a better fashion.

[joe_b]

Fresnel reflection to dark?

I don't understand you, but I know that black painted cars has reflections. I'm sure.
... and my CG cars are so bad looking.

Your right - I was thinking incorrectly. I was thinking the diffuse color was being used as the reflectiveness - this is not correct. My mistake.

[Micha_cg]

Fresnel reflection to dark?

Next chapter: if I set all diffuse colors at a lower value (white is 200,200,200) now) to get physical correct renderings, why not a global faktor, so I can set white and vray works physical correct?

I have done a new test and set the multiplier of the first GI engine at a lower value. Wow, all diffuse colors are darker. Now I can set the color and get it in right brightness in the rendering. In my previous cat test rendering (image5) I have setup a very dark red to get this medium bright red. Now I get what I set. Oh happy day.

image6

[Micha_cg]

Fresnel reflection to dark?

So, and now the simple solution: linear workflow (enable sRGB at framebuffer) and both GI engine multiplier at 0.8. No burn value needed, but a gamma of 0.8 can help to bring more conrast.