[tom]

Are you modeling the contact hydrophobic?

[Bubbaloo]

It's a sphere cut in half. Camera facing side is glass, opp side is mirror with normals flipped. So there is no glass normal exit on the inside. No coplanar faces. Glass IOR is double the mirror IOR.

[Bubbaloo]

Also tried with a full sphere and a micro gap between the sphere and the curved back mirror surface. Same result.

[Bubbaloo]

Eureka! I have stumbled onto the solution. Render coming soon.

[eric nixon]

erm.. I was wrong in the last post about needing refraction for retroref fabric.

I ran another test with just one surface and a normal map of fine indented pyramids, seems to work well, surface alongside just for reference (metallic glossy plastic).



3min render bench 7000, detailed hdri.

[Bubbaloo]

Nice, eric.

On the micro level:

[eric nixon]

er, that previous render isnt working well in subsequent tests,....

Brian, your nearly modelling atoms now? thats not so practical..

[Bubbaloo]

thats not so practical..

Fun is not always practical.

[eric nixon]

Well I'm still confused, but this is result with camra mounted light, 5mins.. might just be some kind of satin rather than true retroreflection.



This is normal map used at 60% with x-flipped

[tom]

You got it, Brian! Now make quite a lot of them....

Eric, you are not reading me. So, again:

It's quite hard to simulate this material accurately using normal/bump mapping. Ideally, it requires to be modeled accurately down to the microns or needs a dedicated shader.

...where "hard" politely means "not possible"... oh

[eric nixon]

Yes I didnt read that post. Oscar once said it should be possible with positive assymetry but maybe he was wrong.

The two surface approach is very difficult to fine tune so its practically impossible to make the ray refract, reflect then refract again so it goes back the way it came in, the normal maps cant indent enough without artifacts, and the ND needs to be correct, AND the gap between the surfaces needs to be just right

[eric nixon]

No, he was right!

Success! Heres 2 images where just the camera has moved, the front panel of the shirt has less than 1mm thickness with an sss based mxm, ND 1.00001, pos assym .8, roughness 7, scattering 6000, attenuation 40nm. NO MAPS USED

1st image; camera is aligned to main light (camera flash)

[Bubbaloo]

Nice!

[tom]

Retroreflection has nothing to do with SSS but still very nice! In other words, you replaced a real phenomenon with a virtual one. It reminds me, "If you can't make it, fake it!" But, OK. This is clever!

[eric nixon]

Yes, and thanks,.... not originally my idea though.. to be fair.