refraction color triples rendertimes

[flipside]

Hi,

I have noticed a very annoying thing while rendering clear plastic/glass.

I made a material:
diffuse=white (255)
reflection= near white, fresnel checked
refraction=white (255)
reflect on backside is turned on

This renders fine.

Now when I turn down the refraction color to 245 instead of 255, rendertimes triple...

Without GI, rendertimes for both materials are nearly the same.

It doesn't matter what I use for secondary bounces, problem remains te same.

In first bounce I have irradiance map. Sample lookup or interpolation type have no effect, the 245 material is always 3 times slower.

When I render the GI part only, the 245 material is about 2 times slower.

The actual rendering part (so without GI calculations) is about 3.5 times slower for the 245 material.

So I thought, I will render the IR map with the 255 material, then save it, load it again and change the material to the 245. That way I can speed up the IR map part by 50%. BUT! Rendering the 245 material with the 255 IR map takes 2.5 times longer than with the 245 IR map! So that's even slower in total.


Why is there such a big difference? Is it that more difficult for vray to tint each pixel a bit with the white of the diffuse?

I can't get the same effect using fog color and 255 for refraction. I need the whitish tint refraction color gives me. By using fog, the material gets tinted darker always.

Probably the client will like the 255 material better, but I don't, and still like to know where the speed difference comes from.

Here's my testscene: http://users.pandora.be/stor1/vray/refraction.zip

The visual difference between the 2 materials is not big in my testscene, but it is in the real scene I want to use it for.

Any help greatly appreciated!

wouter

[vlado]

When your refraction color is white, then the diffuse color is completely dimmed and no GI is computed at all for the glass object.

When you set refraction to 254, some diffuse color is present and VRay must compute GI for the object. When you combine this with internal reflections and back-face reflections, it can mean a lot of additional GI calculations. It may help if you set the diffuse color to something other than pure white.

Best regards,
Vlado

[flipside]

Ok I'll try that!

thanks

[flipside]

Doesn't make muxh of a difference...

Is there a way to disable the GI calculations for that material (like if the refraction was pure white) but still having refractive GI caustics?

wouter

[vlado]

Doesn't make muxh of a difference...

Is there a way to disable the GI calculations for that material (like if the refraction was pure white) but still having refractive GI caustics?

wouter

Yes, make the diffuse color black.

Best regadrs,
Vlado

[flipside]

Mmmm but I don't want that either because then refractions are tinted black when using not pure white for refraction color.

I'm experimenting now with different exit and environment colors to achieve the same effect.

I had a falloff map (perp/parallel) with an output map in one slot (output to 2) to create a self illuminating effect in the plastic, which works quite well but it render way too long.