PDA

View Full Version : Spectral dispersion



OFF
03-31-2007, 07:02 AM
possible obtain such a result with new LW9.2 node system?

Exception
04-01-2007, 10:15 AM
we've run some tests and so far it seems only to do spectral dispersion inside materials. But since I didn't do those tests myself, I'll look into it this afternoon. I think it is possible.

Earl
04-01-2007, 01:59 PM
Yep, to my knowledge LightWave only currently supports the dispersion of light on the shader level of reflections and refractions - it cannot actually cast the dispersed light onto another surface.

Eric Walters
04-01-2007, 02:11 PM
Oh well,

I was hoping to make some "Exceptional" art-like renderings using spectral dispersion...



we've run some tests and so far it seems only to do spectral dispersion inside materials. But since I didn't do those tests myself, I'll look into it this afternoon. I think it is possible.

starbase1
04-02-2007, 05:25 AM
It might be possible in three passes, (for R, G and B), with the index of refraction varied accordingly.

Exception
04-02-2007, 06:47 AM
Weepul said he made it work using a complex shader tree but that it was impractical for common use.
The RGB has been suggested before, I guess that's what weepul did too.

Perhaps in the next version... LW is certainly capable of proper caustics and light dispersion otherwise.

Well. At least your LW render will be done before you retire... :)

Limbus
04-02-2007, 09:22 AM
You could also animate the IOR over 1 frame, use a gradient with the spectral colors and render with motionblur.

Florian

Simon
04-02-2007, 11:48 AM
As Limbus said, there's the old motion blur fake varying the colour and IOR over time. Shame it doesn't work with photoreal mb.

Exception
04-02-2007, 12:04 PM
Shame it doesn't work with photoreal mb.

How come?
did you send that in as a bug?

Nice method btw.

amigo
04-02-2007, 12:40 PM
I would try using proper caustics in a scene setup that has 3 lights: individual R, G and B all overlapping each other and pointing at the same spot.

Simon
04-02-2007, 01:00 PM
Hi Exception - it uses caustics which need the different passes to recalculate each time with the different IOR. Well it does sort of work but you needs loads of passes which sort of defeats the point of photoreal mb.

vbk!!!
04-02-2007, 02:22 PM
Are you sure it doesn't works with photoreal.
if you make your variation between -0.5 and 0 for classic mb. you have to add variation betwen -1 and -.05 aswell for photoreal.
Hope it helps ...

Tesselator
04-02-2007, 02:42 PM
possible obtain such a result with new LW9.2 node system?

Yes. Not a problem at all! Any node with a "Dispersion" setting will do this.
You don't need any "tricks" either.

DonJMyers
04-02-2007, 02:47 PM
I would think what you are looking at are just caustics so radiosity plus caustics plus dispersion nodes are the way to go, eh?

Tesselator
04-02-2007, 02:49 PM
Here's just a white light blasting through a prism.

I'll let others set it up to be more pretty. ;)

Tesselator
04-02-2007, 02:51 PM
I would think what you are looking at are just caustics so radiosity plus caustics plus dispersion nodes are the way to go, eh?


Yeah, you don't need radiosity to get this to happen. It looks prettier with tho. :)

Weepul
04-02-2007, 04:08 PM
Weepul said he made it work using a complex shader tree but that it was impractical for common use.
The RGB has been suggested before, I guess that's what weepul did too.
Nah, I was avoiding the need for motion blur by using a special noise texture to control stuff. It could be done in a more practical way...I just didn't in the case of the simple test I performed. :p


Yes. Not a problem at all! Any node with a "Dispersion" setting will do this.
You don't need any "tricks" either.
Why then, with Monte Carlo radiosity and 100% dispersion on a Dielectric/100% Transparent Standard poly-side-switched material, am I getting this result? (See attachment.)

I'm quite sure that radiosity does not "see" dispersion effects.

Tesselator
04-02-2007, 04:14 PM
Did you set the blur?

Weepul
04-02-2007, 04:20 PM
Yes, and Refraction Blur was enabled.

You can see the dispersion working in the prism where it refracts a little bit of the light on the ground plane.

Tesselator
04-02-2007, 04:21 PM
Try something radical (like attached below) till you see it. Then tone it down
to something normal. Keep in mind that it is actually physically accurate so
your shapes have to be correct as well as your light placement.

T-Light
04-05-2007, 12:11 AM
Thanks for the scene Tess, this is amazing fun :D

7 minutes using interpolated FG :)

jameswillmott
04-05-2007, 01:04 AM
Are you sure it doesn't works with photoreal.
if you make your variation between -0.5 and 0 for classic mb. you have to add variation betwen -1 and -.05 aswell for photoreal.
Hope it helps ...

PhotorealMB doesn't interpolate surface envelopes from what I remember, so you'd have to use multiple passes to get the correct colours coming through.

jameswillmott
04-05-2007, 01:23 AM
Yep, Simon's method (caustics+variable IOR+motion blur) works nicely! You need a heap of blur passes though...

Ztreem
04-06-2007, 05:31 PM
That workaround is several years old, the original tutorial seems to be gone now.

OFF
04-08-2007, 08:56 AM
yes, i remember this, this was founded on light settings in the grapheditor and mot.blur, but, Tesselator say, now (9.2) this possible standard facility. good!

ana027
11-29-2007, 08:23 AM
Here is a test with pure montecarlo 2 bounces.

mdesign
03-13-2009, 02:15 AM
Oh boy! I can't recreate this effect in 9.6. Can anyone give me some instructions, please?