Material news

[jeanphi]

Hi,

I've just pushed a bunch of small material thingies:
- carpaint model now more closely matches the one described in the paper (still no flakes though), I find the results with the presets much nicer
- a new fresnelcolor fresnel texture allows to define the IOR from the normal incidence reflection color (the conversion is meant for metals rather than dielectrics)
- a new fresnelname fresnel texture that will try to load in order a SOPRA file, a Luxpop file or a IOR preset (to mimic what the metal material does)
- a new metal2 material that takes a fresnel texture instead of a name, combined with the fresnelcolor texture, it should open new ways to define metals.

The metal2 material might loose its fresnel parameter and use the interior property instead, I'm not yet sure of what approach is best.

Jeanphi

[Abel]

Very nice! I'm looking forward to testing the car paint. And from what I gather after some IRC discussion, the new metal material will allow metals of any colour, right? I'm sure that will come in handy.

Those SOPRA and Luxpop files seem a bit exotic for my uses but I'm sure the more mystically inclined will be happy to exploit them right away.

[Lord Crc]

Good stuff jeanphi!

Personally I think I like being consistent with glass2 by having the interior define the fresnel component.

[J the Ninja]

Good stuff jeanphi!

Personally I think I like being consistent with glass2 by having the interior define the fresnel component.

This would also make exporter implementation a lot easier, as at that point it is really just a copy of the metal mat code with the preset/file selector removed.

On the other hand, it does seem a little weird having to use the interior volume for something on a material that opaque. I do agree with LordCrc though that consistency with glass2 is probably a good thing. If we are going to add another material that is just an empty shell for another property, it should behave as much as possible like glass2, our other 'empty shell' mat. Users who are already familiar with glass2 should pick this up fairly easily.

Also, on the subject of car paint: here's one of the presets (either BMW 338 or Polaris Silber, now I can't remember which one I set ) rendered with the old code (top) and the new code (below). I've also adjusted the defaults in luxblend25, the new default paint is the third luxball. Also, since this seems to deal with the light loss at angles problem with car paint, I went ahead and simultaneously assigned the bug report to jeanphi and marked it as resolved. http://www.luxrender.net/mantis/view.php?id=1126

Old:

New:

New Blender 2.5 defaults (basically the polaris silber preset with values rounded off a bit and a brightish white base)

[jeanphi]

Personally I think I like being consistent with glass2 by having the interior define the fresnel component.

Me too, however metal materials, contrary to glass materials are often used on "double sided" sheets which have no interior (more precisely they need to have exterior=interior=ambient volume).
That's a bit tricky. Maybe allow a parameter specifying a volume name, then when void it uses the global interior and when defined it uses the specified volume?

Jeanphi

[jeanphi]

Very nice! I'm looking forward to testing the car paint. And from what I gather after some IRC discussion, the new metal material will allow metals of any colour, right? I'm sure that will come in handy.

Exactly

Those SOPRA and Luxpop files seem a bit exotic for my uses but I'm sure the more mystically inclined will be happy to exploit them right away.

They were all already available through the sopra, luxpop and preset fresnel textures, the new texture is just a place holder to eventually replace the code of the metal material to create the correct fresnel texture on the fly and just use metal2. I've not done it yet because I think that metal2 is actually slightly slower than metal, I need to benchmark it.

Jeanphi

[Lord Crc]

Me too, however metal materials, contrary to glass materials are often used on "double sided" sheets which have no interior (more precisely they need to have exterior=interior=ambient volume)

Hmm, wouldn't it then be better to follow the glossytranslucent example and have a "doublesided" switch, which would swap the interior/exterior definitions if it hits the backside?

[jeanphi]

Hmm, wouldn't it then be better to follow the glossytranslucent example and have a "doublesided" switch, which would swap the interior/exterior definitions if it hits the backside?

The thing is that BSDF::GetVolume wouldn't work anymore or we would need to add a BSDF property to tell if we are dealing with a volume or a thin mesh. That would require the user to set that property for all materials.

Jeanphi

[binarycortex]

- a new fresnelcolor fresnel texture allows to define the IOR from the normal incidence reflection color (the conversion is meant for metals rather than dielectrics)
- a new fresnelname fresnel texture that will try to load in order a SOPRA file, a Luxpop file or a IOR preset (to mimic what the metal material does)
- a new metal2 material that takes a fresnel texture instead of a name, combined with the fresnelcolor texture, it should open new ways to define metals.

Jeanphi, could you please provide an lxm with an example of metal2, and if possible also fresnelname and fresnelcolor.

[jeanphi]

Hi,

Here you are:

Code: Select all

Texture "customior" "fresnel" "fresnelcolor" "color color" [0.3 0.5 0.4]
MakeNamedMaterial "custommetal" "string type" ["metal2"] "texture fresnel" ["customior"] "float uroughness" [0.01] "float vroughness" [0.01]

I just found that fresnelname isn't consistent, I'll fix it before posting an example.
The fresnelcolor color parameter can be textured.

Jeanphi