LuxC4D 0.06

[forgeflow]

My work-around for the current lack of surface controls in LuxC4D - I fire up Blender, make my material in LuxBlend, and copy/paste the setup into the .lxm file that LuxC4D exports. Tried this with "carpaint" on a model I had done ages ago:

Tachikoma from Ghost In the Shell Stand Alone Complex

I've been reading up on the carpaint parameters, and realize now that I slightly butchered the setup with my ignorance, physically-wise, but I like the "hot" result and it looks very much like the original shader setup I had on my original C4D rendering. I like the shiney. It rendered really fast too, except for some lingering "fireflies". Well, that's what you guys call em. I call em "snots". I keep wishing there was some kind of tool in the GUI where you could circle problem areas to tell the renderer "concentrate your efforts here". Anyway, still having fun.

[Dade]

Sorry for being totally off-topic but I must say that this model rocks

[forgeflow]

Here's another little project... I wanted to see what it would take to do a "QuickTime VR" of the museum model, so I figured the easiest way would be use an "environment" camera, and convert the image from there. I rendered a 2400 x 1200 image. I think that I over-rendered by a fair bit doing that vs. rendering out 6 different 800 x 800 views, but the setup seemed more straightforward doing it this way. I threw a lot of machines at the problem (4 8-core Mac Pros and 1 8 core machine with hyperthreading running 14 threads as the master), and the render got up to nearly 24,000 samples per pixel when I pulled the plug. There are speckles all over the image, which is not surprising given the image size, the number of light groups, and the number of light sources. I did put portals over the windows, and that did seem to help the overall lighting, but lighting things covered by glass in an enviroment largely lit by bounced light seems to be a tough nut for Lux to crack. I should probably try the display glass as "architectural" as the IOR really doesn't play a large part in the look of the scene, and only seems to exacerbate the difficult lighting situation. Live and learn Next time I'll try the 6 camera approach instead.

Here it is: http://planets.shacknet.nu/images/museumpano.mov. You'll obviously need QuickTime to see the thing, so I'm not sure what people using Linux would need to use...

[tomb]

My work-around for the current lack of surface controls in LuxC4D - I fire up Blender, make my material in LuxBlend, and copy/paste the setup into the .lxm file that LuxC4D exports. Tried this with "carpaint" on a model I had done ages ago:

tachikoma.jpg

I've been reading up on the carpaint parameters, and realize now that I slightly butchered the setup with my ignorance, physically-wise, but I like the "hot" result and it looks very much like the original shader setup I had on my original C4D rendering. I like the shiney. It rendered really fast too, except for some lingering "fireflies". Well, that's what you guys call em. I call em "snots". I keep wishing there was some kind of tool in the GUI where you could circle problem areas to tell the renderer "concentrate your efforts here". Anyway, still having fun.

I agree with Dade - absolutely wonderful model! Nice render too - regarding the values; well, just because it doesn't fit the measured BRDF of the original paper doesn't mean it wouldn't fit the measured BDRF of *some* material in the real world The reason we expose all the parameters in addition to the preset names is exactly for this purpose - so you users can tweak and experiment and make nice renders

At this point, the order of the "lower" substrate layers doesn't matter because lux doesn't support a "true" layered material system yet (ie. with in-order absorption and scattering) so it's basically behaves like a 3 x 1/3 mix-material with a fresnel layer on top for the clear coating (which does have optional absorption by the way, adapted by LordCrc from this paper http://www.cg.tuwien.ac.at/research/pub ... -paper.pdf I think. Also see viewtopic.php?f=10&t=1132&p=9944).

Would be great to see this render in a higher resolution by the way

T

[psor]

I'll join the off topic team. Great model and rendering!




take care
psor

[Abel]

I wanted to see what it would take to do a "QuickTime VR" of the museum model, so I figured the easiest way would be use an "environment" camera, and convert the image from there.

That's great! How did you convert the image to the Quicktime VR format?

[forgeflow]

That's great! How did you convert the image to the Quicktime VR format?

I had to "flip" the image horizontally, as the environment camera produces an image meant to be used as an environment map. Then I used a Mac program called "Cubic Converter" to convert the image to a cubic VR format. Easy when you have the right tools

[forgeflow]

Oy. Here's another render. I was intrigued by the "matte translucent" material, in that it looks a lot like subsurface scattering, which is always useful. So I tried it on the ol' candle wax rendering. This has some very basic, unrealistic modelling, as I was just trying out the idea without thought of actually making a composed rendering. What I found is that on it's own, the matte translucent material is TOO translucent to give a good SSS effect. What you have to do is put a black object inside the translucent object to soak up the excess light, and give it some mass. Works perfectly. So here you go, Merry Christmas:

faking subsurface scattering

I should probably try putting a skull into a head object to see if it works for skin too.

[Lord Crc]

Very nice! Good idea about the "inner core" as well, worked out very well.

[jeanphi]

Hi,

Actually you don't need the core to be black. To better fake SSS, the core reflection should approximation backward scattering and the core transmission should approximate forward scattering (factor in the absorption as there's currently no way to do real absorption).

Jeanphi