[Thomas An.]

the first dispersion doesn't pass through the second prism

The things to double check are:
-Surface normals (all normals need to face outwards)
-The ND and abbe values. Make sure it is ND=1.847 and abbe=23.8.
-Also each prism in the DXF is made of two parts, select the portion that includes the top, bottom and back faces and make it black on each prism; (use the same black material as the the one on the light box to prevent infinite internal reflections and hopefully speed up the process).

I see your beam escapes the same as the red one from "thedonmakaveli" on page 5, but he fixed it (I suspect the surface normals)

[def4d]

ok i'm retrying with only one prism
I've found an interesting camera view that shows the full dispersion from the very first sample!!!
I'll post a decent render in a few hours

[def4d]

So this it



2H44 @ 13.67 Samples

The dispersion has been immediatly visible at sample 0.00 in the prism, with the same intensity , though it took 13.67 at the intensity you see for the wall

Notice the bending too

[tom]

Very nice

[def4d]

Thanx Tom, but have you got any explanation on why it appears so fast in the prism, and not on the wall?

[tom]

Not exactly, but it seems to me a magnifed internal reflection of the dispersion pattern formed after the first refraction route which needs a very low angle to be observed by the viewer. You're lucky So the energy solution would be faster than the one on the wall, Oscar Cano is the one who would be able to answer this properly.

[def4d]

Thx Tom
And i know how much i'm lucky: i've changed the ior a bit, and pfoooo... no more visible internal dispersion
Via Thomas test, i 've found a very interesting search field in the Optical phenomenas (hope to see him again!)

[tom]

It would be cool if you keep and test this scene with 1.0 also

[def4d]

Another one, for wih i've added a lens front of the gun (just to see)
Just that the dispersion in the prism has no decal, contrary to the previous

[breberka]

im sorry but i have no time to read it detailed. may be a question on the bending dispersion effect. how is your emittor body material definition. is it absolutely black color, absorbing everything? you need to simulate only a small gap with emitting photons, these photons have to fly collinear. i would guess ... if you have a small deviation, through a numerical errors in a maxwell solver and reflections in a prismas, you could have such an effect. but i don't now which numerical method maxwell uses, i guess finite volumes or stg like this. it should be nice to see energy convergency. but if you rendered app 1000 hours ... omg.....

next point.... i would say that maxwell is based on some equations. these equations trying to describe real world ... there is some small error in an incomplete equation development, but for some reasons we are not able to solve more complicated equations. you have to add some small error comming from numerical description of maxwell equations and may be some rounding error of your processor.

so my conclusion ... something wrong in a definition ... or it is a result of a sum of a small errors given by methodology. it would be nice to have an energetical balance ... means energy of emmited light versus energy of a light on the wall.

hmmm. im getting tired ... i guess i ll read this thread complete. and make some tests....

[breberka]

tom: Superimposed in Photoshop

hihi, very nice sentence.

[def4d]

tom: Superimposed in Photoshop

hihi, very nice sentence.

[tom]

tom: Superimposed in Photoshop

hihi, very nice sentence.

...don't worry, this could also prove that you are physically correct, too.

[def4d]

and what does it give, in an explicit way?

[Thomas An.]

im sorry but i have no time to read it detailed. may be a question on the bending dispersion effect. how is your emittor body material definition. is it absolutely black color, absorbing everything? you need to simulate only a small gap with emitting photons, these photons have to fly collinear. i would guess ... if you have a small deviation, through a numerical errors in a maxwell solver and reflections in a prismas, you could have such an effect. but i don't now which numerical method maxwell uses, i guess finite volumes or stg like this. it should be nice to see energy convergency. but if you rendered app 1000 hours ... omg.....

next point.... i would say that maxwell is based on some equations. these equations trying to describe real world ... there is some small error in an incomplete equation development, but for some reasons we are not able to solve more complicated equations. you have to add some small error comming from numerical description of maxwell equations and may be some rounding error of your processor.

so my conclusion ... something wrong in a definition ... or it is a result of a sum of a small errors given by methodology. it would be nice to have an energetical balance ... means energy of emmited light versus energy of a light on the wall.

hmmm. im getting tired ... i guess i ll read this thread complete. and make some tests....

Thank you for this insigt breberka !
You sound like a developer

Yes the emitter body is absolutely black HSV=[0,0,0] ... actually it doesn't matter, because I tried other values too, but the bending is still present.