[lebbeus]

The power went out last night during my test

I'll get back to this later today

[Tim Ellis]

The trick for retroreflection is to have a complete sphere halfway embeded in some opaque material. This way, no matter what angle a beam of light hits it, it will refract , reflect, and refract once again - exiting the bead at exactly the oposite direction it initially entered.

Here's an example on how it's used on street lines.



So Ian - after you generate your trillion spheres, you'll need to duplicate them, boolian the duplicates from a thin box, then bring back your original trillion spheres with a glass shader. Should work fine then....

This is what I had in mind originally. The 1/2 spheres in the file I posted, are scaled very slightly larger than the spheres, so they can be aligned to work as Michael posted, without intersecting or having faces sharing the same space. The face normals are all calculated to point in the correct directions.

I seperated them to give more options to anyone who uses them. (That is if they work! I've not had chance to test this yet.)

Tim.

[lebbeus]

From what I was reading in the Wiki there are two ways to accomplish this, one with the metallic half sphere and glass (or the partial prisms), the other with a dielectric that was exactly twice the IOR of the incident medium. Assuming that Maxwell considers air which (for the significant digits that we're able to use) would be an IOR of 1.00 we'd only need spheres that had an IOR of 2.

I tried an expereiment with spheres that had an IOR of 2 versus spheres that had an IOR of 1.55---there was really no difference though I only let them cook for 2hrs. [Of course I've probably set it up incorrectly…(i'll post the images tonight)]

It'd be really interesting if there was a complex IOR file for these types of images available somewhere on the web. I checked 3M's site and some others but they didn't have any available. How does one go about creating an IOR file? Has there been a thread about that here (haven't done a search yet)?

I don't have any doubts that this could be done with a brute-force geometry solution, based on the other experiments that people have done with Maxwell--however this wouldn't be practical or useful for anything other than playing around. Maybe it can't be done any other way yet, I don't know…it's just a crazy idea that popped in my head.

I appreciate everyone's interest in this.

[lebbeus]

random links that may or may not help with this "distraction"

http://swiss.csail.mit.edu/~jaffer/Free ... l_toc.html
http://mrr.nrl.navy.mil/pubs/Design.pdf
http://www.esrf.fr/conferences/proceedi ... rie99b.pdf
http://www.opticsexpress.org/abstract.cfm?id=75148

[Tim Ellis]

My first tests.




The camera is 50 meters away from the reflectors.
The reflectors are 28 mm x 64 mm of each type.
Light source is just behind the camera & 10000W @ 100%.
20 minute test render.
No sky/skydome/mxi/hdri.
Camera zoom @ 3500mm.

I've left one running over night of the spheres & cups, as they came out the best.

Tim.

[lebbeus]

been thinking: maybe the retroreflector material is closely related to a car paint material. They both have "flakes" that reflect light, only the retroreflector is more coherent, rather uni-directional in its reflection.

will test this…

[lebbeus]

tests are still going, but I've been finding some interesting things online:

Coefficient of Retroreflected Luminance (RL): A measure of retroreflection most often used to describe the retroreflectivity of pavement markings. Coefficient of retroreflected luminance is defined as the coefficient of luminous intensity per unit area.

Coefficient of Retroreflection (RA): A measure of retroreflection used more often to refer to the retroreflectivity of highway signs. Coefficient of retroreflection is defined as the ratio of the coefficient of luminous intensity (RI) of a plane retroreflecting surface to its area (A), expressed in candelas per lux per square meter.

[lebbeus]

I've been messing around with the material system and I think I've come up with a solution, or at least a direction for further study.

I've used Mverta's sample scene.

This is a two layer bsdf.

first layer: lambertian diffuse black material

second layer: "glass" with IOR of two, 0 roughness, and a random noise map (see car paint discussion:http://www.maxwellrender.com/forum/view ... hp?t=16330 ) in the normal map slot with a very thick coating (10500nm, Nd1.00).

the normal map is a bit screwy, but it seems like I'm getting the kind of reflection I'm looking for in a few areas. I'm thinking that if the normal map gets worked out correctly (somehow approximating micro-spheres on the surface) then this material will be "solved".

[jfrancis]

Consider two mirrors at 90 degree angles to each other -- more specifically, try a 90-degree-angle cone.

It's like a billiard ball with two banks. Hit the first bank shallow and the second bank will be steep and the ball will exit the corner on a path parallel to the one on which it entered. No matter what angle the ball enters the corner it will (barring spin) return on a parallel path.

The Apollo astronauts left an array of conical mirror retroreflectors on the moon that they use to bounce lasers from earth to moon to earth again.

http://www.truemirror.com/moredata.asp

http://farside.ph.utexas.edu/teaching/3 ... de127.html

[Cadhorn]

Bump... almost two years later.

One of the recent "sense" contest winners, the material "reflector", reminded me of this thread on retroreflective material.

One of the hold-ups seemed to be that normal mapping was broken in a previous version of maxwell. now that that's fixed...

Did anyone make any headway on this?



[This whole thread should be moved to the MXM section... mods?]