PDA

View Full Version : two transparency channels



habaņero
11-01-2005, 10:05 PM
I would like to have an optional additional transparency channel that would eat at the primary one, eg zilch it out. It gotta have at least incidence angle preferrably full controls.

Pretty Please, it is for internal reflection with Fprime's caustics;

http://forums.cgsociety.org/showthread.php?p=2787435#post2787435

Dodgy
11-02-2005, 03:59 AM
How is this different from adding another transparency layer?

habaņero
11-02-2005, 08:03 AM
It is a bit hard to explain, check out the images in the linked thread.

At 100% transparency, If you set refraction at .001, the rays will bounce back. And they will produce great caustics with kuminos objects or HDRIs (no lights), that you can't get with normal reflections. You can even magnify the caustic effect witout burning out the reflection.

But then you can't also have it to show transparency, and so you can only get refracted caustics with transparent objects. Putting a fresnel on the RI, it will not produce a gradient but a cutoff. And with alpha on this gradient, it takes the middle RI value, not a mix. 10% transparency and 90% percent diffuse though works perfect, you can have fresnel on solid objects.

I managed to get some results by using textures on the RI, but not very usable ones ...

habaņero
11-02-2005, 11:59 AM
Obviously, an additional reflection channel called "Fprime caustics reflection" "Alternative reflection calculation"or similar could be a better name for the feature, if it would be possible to do it that way ...

Captain Obvious
11-02-2005, 03:08 PM
This isn't a Lightwave feature request. Lightwave's radiosity takes reflective surfaces in consideration. Ask Worley to make Fprime consider not only transparent but also reflective surfaces when it samples the global illumination.

Captain Obvious
11-02-2005, 03:28 PM
Pardon the lousy quality in these two renderings. It kind of sucks doing test renderings on a 1.33GHz G4... ;)

The scene consists of a ring that is 100% reflective, and 0% diffuse and transparent. Raytraced reflections are turned on, otherwise this does not work. The first image is rendered using a highly luminous polygon, and monte carlo radiosity (one bounce). If you look closely, and try to see through the noise, you can see the same pattern as in the second image, which is rendered using a point light and regular caustics instead of radiosity. In order to get this effect in Fprime, you need to make it transparent and set the IOR to a very low value, as far as I've understood it. This is not needed in F9, so it is obviously a flaw in Fprime, not in Lightwave.

habaņero
11-02-2005, 04:10 PM
Well I mailed Worley this thread, so it is well possible your suggested solution will reach him.

With the new nodal texture system, I reason it would probably be an opening for this anyway, but just in case I would like to mention this to the developers since a little thing like this could mean a difference to something as beneficial to LW as Fprime.

I don't think it is a deficiency in either Fprime or Lightwave, it is a priority on Worleys side I guess. But if you don't believe this would hurt lightwave, I hope you can accept that I personally still would like this option, if it isn't unreasonably hard to accomplish.