[Ernesto]

I have read in the manual the technical explanation of K.
But I wonder when to use it?

E

[Bubbaloo]

Used for metals.

[Ernesto]

I see...
Does it affect reflections?

E

[rickbern]

Ernesto,

I'm not the world's ranking expert here, more that I don't know than I do, but I think you're usually better off using measured ior data than using K. I think K gives a more metallic look at a lower IOR than if you don't use it. I don't think using measured ior's really slows down rendering that much.

I find that to make an alloy, I add a separate bdsf for each component of the alloy and then set the percentage to match the real composition of the alloy I'm trying to create. Works pretty well.

While we're on the subject, if anyone knows where to find measured ior for tin I'd love to find it. I'd love to try to make actual bronze alloys, see how they work out. It's bizarrely not in the ior database and I can't find the data online.

Rick

[bograt]

I agree with rickbern, but it is rarely that I want to render lab grade metals.
I always have to deviate slightly from the original wizard material to achieve realism, but as stated in the user manual, K is usually not neccessary to achieve the desired result, rather it adds control to complex fresnel equations.
For you Ernesto, like me, it might not be worth the trouble of changing.
Also you should print this out and read it on the train or something, it's actually a surprisingly good read :http://www.maxwellrender.com/pdf/maxwel ... .5_eng.pdf

[bograt]

(I mean changing the k value, not yourself)

[Half Life]

K is pretty easy to understand --it is the correct solution to the old way of making metals. Meaning in version 1.x if you wanted to make the metal more reflective you would increase the Nd, which is not correct, and leads to artifacts.

By using K instead you can eliminate the "dark edge" artifacts as seen in this step by step example (look at each MXM in order and you will see).

https://dl.dropbox.com/u/41250852/metal1.mxm
https://dl.dropbox.com/u/41250852/metal2.mxm
https://dl.dropbox.com/u/41250852/metal3.mxm

Best,
Jason.

[bograt]

Good point

[tom]

It's true. K is there to make correct Fresnel curve for non-dielectric reflections. When K=0, Fresnel curve has a cavity close to the grazing angle called Brewster's Angle. This is correct for glass, plastics etc. but metals. Metallic reflections require K>0 to avoid this darkening on the edge.

[Ernesto]

Thanks you all guys!
I understood the dark edge issue, and agree that getting rid of it metals look much more realistic!
The phisical explanation looks prety complex to me, but I have got the concept.
I can see that the K value could go from 0,01 to 999, and as set by the Materials Wizard it goes arround 3 or 7.
Now that I know what to look at, I made several tests and I could see a very subtle effect.
The lower K number it seems to increase the dark border effect. Am I right?

E