[thedonmakaveli]

Thanx Thomas.
So instead of a box i must do a cilinder?and it's material is important or justa diffuse is good?
The shape of the emmitter it's good or i must change it?

[Thomas An.]

No, No. Remove the blue box completely. (do not add anything to the file I gave you)

Also, (just making sure) the white thin box is not an emitter. Make that a black diffuse.
The thin surface on the back of the white box is the emitter. Give that a 10,000w strength. (and make sure the normals are facing towards the prism)

[thedonmakaveli]

Sorry, maybe i'm very stupid...
But the thin line (yellow one) under the box it's just a polyline that goes thru the all scene.. i must extrude it and cut it ?
here the perspective...
The yellow line it's the light tajectory i think....
Sorry i know im makin you crazy... thanx 4 your help...

[Thomas An.]

[thedonmakaveli]

Did what U told me....and now i tried with red light because u said that's faster because it's monolight....but in 1 hour..nothin but this...

[Thomas An.]

I explicity said to use D65. (once you do, let it go for 6 hours or more)

Never use monochromatic light for dispersion. Always use white light, but beware that it takes a long time (there is nothing you can do about it, just use white and wait).

[thedonmakaveli]

Yes i Know..i just tried with red light (monolight) just to see if trajectory it's ok...
But now i changed to d65 and i set render to 8 hours...
thanx...
See u later for upgrades.....

[Thomas An.]

ok, so far so good.

You should set it for some wierd number like 9000 hours; this way you can stop it sooner if you want. If you set to only 8 then you restrict yourself.
(Ideally you should let it run up to level 21... for the full intensity of the caustics to be relatively decent)

Another thing: Make sure all glass surface normals are facing outwrards

Let me know how it goes

[thedonmakaveli]

OK
now it's workin'!!!
Another question, how did u do the render with wireframes??(like illustrate & toon)..????
it's very nice...

[Thomas An.]

OK
now it's workin'!!!

Great! Show us your results

Another question, how did u do the render with wireframes??(like illustrate & toon)..????

Superimposed in Photoshop

[thedonmakaveli]

Thankx to u finally this is the result....
Did u used a plugin for photoshop??

[Thomas An.]

Yup, you got it and thanks for posting this test.

Also, like I mentioned earlier, the arched caustics are still present even in this simplified version of the scene.

So, the original question is still there: how can a straight emitter produce curved caustics....

[thedonmakaveli]

Maybe it's the fact that we have 2 prisms... in this other Tom's experiment we have just one prism....
But why he used a lense? and what's material and the composition of this lense?.....

[Thomas An.]

Well, that is a completely different experiment, because the geometry of the emitter is a little square (or circle) but not a long stripe. So the dispersion result is just a series of blurry round dots.

The lens does not do anything to the shape of the caustic. It only adjusts how blurry they are on the wall.

Again, I see no reason why the caustics should bend instead of being straight.

and what's material and the composition of this lense?

I think he used a low abbe (like 2.0 or 3.0), but that is unnaturally low.
http://www.maxwellrender.com/forum/viewtopic.php?t=4170

[def4d]

Thomas : In real life, does your dispersion of dispersion produces this bending?
Didn't find an exemple on net

This is my result with your dxf (had a lot of inverted faces?!)
And i certainly moved something, cause the first dispersion doesn't pass through the second prism