New glossy translucent material

[jeanphi]

Hi,

You have to multiply f by cos(wo.n) and divide by 2pi to get a correct result (look at BxDF::rho to see another way to get the result using the BxDF sampling to reduce variance).

Jeanphi

[jeanphi]

Hi,

Here is the correct archive with my version.

glossytranslucent.zip

(17.78 KiB) Downloaded 25 times

Jeanphi

[jeanphi]

Hi,

Here is the modified LuxBlend to help you try it out.

LuxBlend_0.1.zip

(104.57 KiB) Downloaded 14 times

Jeanphi

[jensverwiebe]

Hello

Just to say that i incuded now glossytranslucent in recent OSX weeklies for better feedback.
( the exporter is a fully patched 0.8 version, so if anybody needs it for other OS, you can DL the zip and just use the exporter.py )
Generally it seems to work fine, but i havn´t done deeper testing yet.

Great job folks.

Jens

[paco]

I've had a chance to partially test the latest version:

The leaves look similar. The hemispheres though look a bit different - the transmitted light seems more blurred in the new version. I Haven't had time to really look at the code - may be more correct than the previous version but i'm not sure.

Also a personal preference of mine: I find it much easier to set the diffuse texture =reflection texture and then just modify a transmission factor 0-1, rather than having to adjust both the transmission and reflection values to keep it physically correct (ie, R+T<=1). For instance in the hemispheres above I had originally set the diffuse to about 0.56 and then just modifed the transmission (0.0, 0.5, 1.0). With the current version I had to modify the diffuse term for each one (0.56, 0.28, 0.0) as well as altering the transmission(0.0, 0.28, 0.56).

[vimax]

Actually the leaves don't look at all the same to me, especially switching between them side by side:

Original vs Correct

Is that the closest match you could find?

Edit: fixed link

[jeanphi]

Also a personal preference of mine: I find it much easier to set the diffuse texture =reflection texture and then just modify a transmission factor 0-1, rather than having to adjust both the transmission and reflection values to keep it physically correct (ie, R+T<=1). For instance in the hemispheres above I had originally set the diffuse to about 0.56 and then just modifed the transmission (0.0, 0.5, 1.0). With the current version I had to modify the diffuse term for each one (0.56, 0.28, 0.0) as well as altering the transmission(0.0, 0.28, 0.56).

Actually you don't need to have R+T<1 at all since the formula for transmission is (1-R)*T, so you just need to leave the R value constant and change the T value. Actually T models the absorption when going through the surface. This will behave differently than your model because when you increase T the object will become brighter (more light passes through but the same amount is reflected).
Does that make sense or should we change the parameters of the model (the current scheme is the same than mattetranslucent with energyconserving=true)?
Thanks for testing.

Jeanphi

[Conz]

Actually the leaves don't look at all the same to me, especially switching between them side by side:

Original vs Correct

Is that the closest match you could find?

Both links are pointing to the same image. Could you fix the wrong one?

[paco]

The leaves were a very quick trial - I'm sure it could be better matched - will have another go when I have more time.

jeanphi - I see what your saying. In my model there was the specular reflector, then anything that got past that was reflected/transmitted according to the transmit variable. In your model anything that gets past the specular AND the diffuse reflector then gets transmitted according to the transmit value. Hmm... I guess they are equivalent mathematically - just a matter of which is easier to conceptualise.

For me at least it seems easier to think in terms of "this object transmits 40% of the light" (roughly) rather than "this object transmits (1-R)*T " of the light. But then again keeping material parameters consistent also makes a lot of sense. You've got a lot more experience with this than me - so i'm happy with whatever you want to do.

I'm curious about the blurring of the transmitted light though. Don't really see where that comes from. I'll check my scene settings again when I get a chance - it's probably a bug there. If it's a real phenomenon then it looks suspiciously like SSS

[jeanphi]

Hi,

I decided to move forward and commit the stuff. Let's work from there and improve the current scheme if needed.

Jeanphi