Well, it should spit out a number between 0 and 1 (0% -100%) which either is the amount reflected or transmitted. So far IOR from medium A and B and the incidence angle should be the only input parameters I can think of.
Right now I do use Excel to calculate these settings which I then convert into a gradient. Pretty boring and if you need to come up with new calculations it takes 10-15 minutes to set everything up.
It should use the original fresnel equations. From my point this should be super simple to set up for somebody who knows how it works.
Gradients only work with incidence as a input parameter, but that doesn't help at all.
Yes It should be possible to have inside and outside IOR,
this current beta take 1.0 for air (outside).
Yes it take both s- and p-polarisation into account.
This algorithm was originaly designed for water effect,
but need to be more tested with Lightwave.
Hope to make an update soon.
Added a note here, that could be discussed:
May be reflection output should not be inverted
like it is in the current beta fresnel node,
But have found some glass material which take
both Transmission & reflection from reflection
output in the equation (inverted for Trans.)
with this comment...
"Physically incorrect, but portable"
Hmm, not sure if I understand what you mean. If there is a "non-physical approach", I'd like something like a checkbox to switch between correct and "in-correct" if it's possible. Normaly I would use this node to change opacity in a mixer node, with reflection and refraction as the input. Did you think of something else?
Yes me too,
Not the Fresnel Node itself,
but a node tree with it for glass with
raytrace functions !?
Use the Spot Info to get Object World point position,
and Ray vectors from Fresnel Node, for both reflection (blurry?)
and refraction, use the transmission and reflection scalar
for transparency and reflection, filter also specular and some other things...
Great work! thank you
Ok I've set up a network that seem to work, but the render is taking forever...
I'm testing your fresnel against the old fashion'd "real-fresnel/IOR/air-polys" setup.
I'm taking advantage of the "polygon-side" feature to get rid of those air polys, in the node version.
Anyway, i dont understand what you mean by "Object World point position,
and Ray vectors from Fresnel Node, for both reflection.."
left thing is node-made with the Fresnel node.
To the right is the classic flipped-air-polys/Real-Fresnel.
The air polys are displaced 10um to avoid artifacts.
The diamond is node-surfaced, with the same network, set to 2.6 IOR.
Refraction on the diamond seems to work also.
(i avoided blurs/antialias/aberrations by now to speed up renders)
It seems that the node version works even better...
(But i still don't get what that world spot thing really means...)
This sample is convincing,
"Object Spot" is the position of (spot) point, being evaluated
in object coordinates, "World Spot" is its position in world coordinates,
this one could be the first parameter of a raytrace node,
the other parameter is the direction vector of the ray.
Thank you denis, i think i understad that,
what i don't understand is how to use that vector info with a raytrace node... or what could be the advantage of that.
Anyway it's not your job to teach me that, i'll try to guess it for myself.
You did a great job, the fresnel node is ok,
and i don't know if my network is the best one could build but for now is working, i'll keep improving it.
when i make something more worthwile than 2 cilinders and a diamond primitive on an infinite plane i'll post it here.
thank you again for making life a little better for us lightwavers.
In fact raytrace function should do almost the same job for
reflection and refraction than the specific nodes you have used,
May be the result you get should be scaled with Fresnel output
Reflection and Transmission scalars, but not sure that is necessary.