Maxwell Render | Official public forum

2 tests:

4 Emitters in different colors to the left
3 lenses, a wall and an imaging plane.
You can also see the reflection of the emitters in the lenses
Rendertime 1 hour and sl 14.45, so the noise-issue is just a question of letting it render overnight.
Again a simple setup and a nice result.


3 different demonstrations of an optical design for a "lightline"
Each one has a LED at the right end.
The Green one consist only of the white hollow cover.
The Red one adds a lightguide within the cover, but without any optical
features, therefor no outcoupling of light.
The blue one is similar to the red, but with optical features added along the lightguide.
Theese features produce the nice even lightdistribution along the entire length.
Rendertime 4 hours and sl 14.59

I actually have real moulded parts of the blue version and it works just fine.

BR Dane

yes, it would be very interesting to see the optical features..

No Dane, tell us you're not serious, because you know some other people think Maxwell is not physically correct
Perfect tests as usual !!! Keep us posted
You are master of optics...

DELETED

O well.. i still like the tests Dane do, they're very interesting to look at.

Nice work Dane!

/ Max

whiskey wrote:as long as it depends on polygons it is useless for complex optics..sorry tom

You are right, but it's not about being physically correct, it's about accuracy.

Lets just say physically attractive

Great tests again Dane! I don´t understand how the lightline scene is setup though....

Mihai Iliuta wrote:Lets just say physically attractive

probably...

Let's make clear these two statements:
a - Physically correct
b - Physically accurate (as said by some others)

[a] is a correct statement because something can be physically correct or not, this is not about accuracy.

is not so correct because here accuracy means "ability to be close to perfectness", so accuracy may change forever. anything can be more accurate than anything. this doesn't mean maxwell is not accurate, it is as accurate as the samples you render. it's the matter of being more precise...

p.s: forgive me if i'm wrong about English.

But Maxwell should really have the ability for parametric surfaces/objects.....why not even provide users with a ready made library of lenses for example? Just using polygons wastes a lot of the accuracy that Maxwell is capable of.

You're right Mihai, this is not even 1.0, think future

whiskey wrote:

other people think Maxwell is not physically correct

as long as it depends on polygons it is useless for complex optics..sorry tom

Unless we had ENOUGH polygons. How many polys would it take to simulate a 1/4 wave accurate telescope mirror? (No, I'm NOT serious.)

I don´t know....but I´m inclined to think that rendering with a parametric lens would only affect the bending of rays to make it more accurate but not necessarily increase the amount of rays (unless you wanted), so it shouldn`t mean much higher rendertimes.

For example like those dedicated optics raytracers whiskey and Dane showed pics of. They shoot a rather small nr of rays but do so very accurately. At least that´s what it looked like.

DELETED

I feel strangely inclined to change my avatar to Mr Burns.......eeeexcellent...

Hi All

Thank you for your interest in the combination : Maxwell and optics.

Regarding the optical simualtions (raytracing) of which I have shown examples in a previous post :
Actually I most often use around 100000 rays in order to design and optimize a system. The final raytrace will use 500000 to 1000000 rays, which is sufficient to get a valid result ... if the input and lightsource is well defined and matches the real world closely.
I also have the option of approximating the geomtry with polygons, which in many cases speed up the raytrace, but with a loss in accuracy... so one has to be carefull.
Also there is a big difference between imaging and non-imaging optics. Imaging being the most critical when it comes to creating complex surfaces. Non-imaging optics is however often in many cases a compromises between nice optics and production facilities - moulding, costs etc... Therefor it is not always as critical.

Regading my lightline.
I can not show you the lightguide with the features, due to confidentiality reasons, but below the lightguide without the features.

The red surface is where the light exits. The design only works correctly if the lightguide is inside the white cover. I can tell you so much that the features are not on the opposite side of the exit surface.

For me Maxwell is a tool for visualizing optical non-imaging designs and as such the accuracy is good.
And the similarity between my technical simulations and Maxwell validates, at least to me, the use of Maxwell for this purpose. I wouldn't however not use Maxwell as the only tool ... not yet anyway.

Also right now Maxwell doesn't allow us to acces the photometric values, but I hope this will be implemented soon.

BR Ole