withe diamond

Here is an example of the quality that I'd like to have with the rendering of gemstones.

Although Maxwell is an excellent high-end renderer, I think you can easily reach similar/comparable results in Cheetah for this kind of scene... with the right setup/hdri/settings, of course:smile: maybe caustics (which I don't see in your example image, btw) will be a little less easier compared to other biased/unbiased renderers, at least in my experience (but it may as well be me and not Cheetah :redface: )

happy rendering!
Alessandro
 
hello

I work lot of time ... to find the best result for the rendering of a white diamond.

Colored diamonds are much easier to make attractive

Cheetah is currently unable to apply an image HDRI affecting isolated material.
HDRI image affects the entire scene. For my needs, it's bad, because I'd like:

Metal => cold and not too rich reflection
diamond => the opposite .. rich, especially with "fire", as in reality.

So, I cheated ... :) And I applied a texture type "checkerboard" inside the diamond, for the "fire".
I have also seen that the specular settings improves rendering.

Wat dou you think and how improve my quest for perfect diamond? :smile:

I attached the material "diamond white" and 3d model for your test if you wish. ;)

files.me.com/dayer/mvnjbe

Good day.

Antoine
 

Attachments

  • diamant blanc.4.2.c3dmat.zip
    18.2 KB · Views: 394
  • test.jpg
    test.jpg
    130.8 KB · Views: 675
  • j12.jpg
    j12.jpg
    143.1 KB · Views: 764
Last edited:
Hi Antoine!
don't know exactly what you consider a perfect (or very well done) diamond buy I agree with Andrew you made an excellent work with this project and improvement over the first shots is stunning... well done!:icon_thumbup:

maybe you could try different HDRIs (with some colors too) to see if you can find the "perfect" one for you

and thanks for sharing your stuff BTW

cheers,
Alessandro
 
It might simply be a question of building out a scene with fake soft boxes and light sources rather than trying to get an HDRI to solve the problem.

Alternatively, render the scene with two different settings (one for diamonds and one for everything else) then render a masking layer and composite.
 
right Podperson,
of course a good studio setup may be better (I was thinking of a good studio hdri indeed, with some "faked" studio light like soft boxes).

the real problem here, if we wanna go beyond the already good shot Antoine did, is light dispersion properties in Cheetah3D. Maybe Martin or Frank can help here, since I suck at this kind of stuff (caustics too) in C3D... I'm finding it way simpler to achieve in physically based renderers (given you have hours for waiting for a decent shot)... I'd desperately need a tutorial on this

cheers,
Alessandro
 
hello all
many, thanks for your feedback.:icon_thumbup:

maybe you could try different HDRIs (with some colors too) to see if you can find the "perfect" one for you

no, no because:

Cheetah is currently unable to apply an image HDRI affecting isolated material.
HDRI image affects the entire scene. For my needs, it's bad, because I'd like:

Metal => cold and not too rich reflection
diamond => the opposite .. rich, especially with "fire", as in reality.

---

It might simply be a question of building out a scene with fake soft boxes and light sources rather than trying to get an HDRI to solve the problem.

Alternatively, render the scene with two different settings (one for diamonds and one for everything else) then render a masking layer and composite.

oulala... it is complicated this?

regards.

Antoine
 
Alessandro:

I agree viz physical rendering. I think that when you start dealing with really complex lighting effects, you're better off using a physically based renderer (unless rendering time is your dominant issue -- e.g. rendering animation)

Luxrender is free and I believe there's even an exporter from C3D. Worst case you can export to blender and then use Lux.

I've never gotten caustics to do what I want/expect for anything beyond the most simple cases, and then it's always a hack. Why bother? Save hours of tweaking and just render it overnight in Lux :)

Antoine:

So as to my last suggestion: render twice, then make a copy of your scene and make a white (flat color) material for your diamonds and a black (flat color) material for everything else and render. Use that image is a mask to composite the first two renders. Not complex, but a little time-consuming.

As to my second last suggestion: building out a scene with soft boxes and lights is going to be a "gift that keeps on giving" -- you'll learn about lighting, and you'll produce a scene that you can re-use over and over. As a bonus, it will possibly run faster, probably produce better results, and definitely give you more control than HDRIs.
 
Last edited:
I found something interesting in reality:

- Real white diamonds photographed by a photographer have NEVER "fire" colored.

- Real white diamonds in a lighted showcase ALWAYS have "fire" colored.

interesting no? maybe halogen source in the showcases..?

therefore it is necessary to add a subtle color IN the material of the diamond.

--

I found this material "colorglass2" on the forum, which gives good results, what do you think?
but many many many node inside ...:eek:
 

Attachments

  • colorglass2.c3dmat.zip
    20.3 KB · Views: 334
  • j12.jpg
    j12.jpg
    137.4 KB · Views: 734
Hi!
Antoine, the model you provided was so cool I had to do some test with this "diamond material" thingy (first time on diamonds and no expert here, so take everything I'm saying/showing with a grain of salt - mine are questions and not answers/software comparisons).

1st shot is a Cheetah3D render with a studio hdri, a low-power area light and a point light for caustic photons only (and you see some caustic "noise" too). Meterials are def Silver with some roughness and anisotropy, Archie's colorglass2 for diamonds (looks more shiny and "alive" to my eyes than the def Diamond material). render time about 20mins at 800x600 (CPU: core2duo@2,4GHz). - note that with def diamond render took less than 7mins

2nd shot is Octane, a studio hdri, diamond material is just a "specular" (i.e. Glass) with proper IoR (2.417) - no dispersion available at this stage of dev (it's the first beta running on OSX). some "camera" adjustment like exposure, real film response, DoF etc. render time about 2hours at 1024x512 with pathtracing and maxdepth=16 (done with a very poor GPU as you can see in the bottom... i think this would need no more than 20mins on a decent GTX2xx) - note that this is a screenshot since I don't have a license and can't save scenes/renders and the statistics at the bottom/left of the page are kind of broken (26 Msamples/s :eek: ) except samples/pixel info

3rd shot is LuxRender, a "real world" hdri, diamond material is a "glass" with IoR=2.378 and CauchyB=0.01415 for dispersion (data grabbed from the Lux forum - user/developer Jeanphi). some tonemapping and lens effects (Glare, slight Vignetting and Chromatic Aberration... don't change that much to diamonds appearance). render time... about 20h at 800x600 (CPU: core2duo@2,4GHz).

Now, none of the diamond are really convincing to me (probably due to "operator error", or noobiness in this case) but it seems the only way to get the "fire" is having a dispersion property for the material. (Cheetah and Octane have no "fire", Lux a bit too much imho maybe due to hdri - I have no studio hdri in exr format, sorry)
On the other hand, consider Octane and Lux shots are almost "out of the box" since I'm very noob about their material settings, while in Cheetah I made some more tests (not much anyway).
I'd be very curious to test the new volumetric glass material in Lux 0.7RC1 but I have serious memory problems with it at the moment (I used the "old" stable 0.61 here).
Of course Lux is prohibitive when it comes to render times (unless you have a superCPU or a render farm) while Cheetah and Octane are pretty quick.

so, since the main reason for this kilometric post was to learn and improve skills in Cheetah's diamond renders, any idea/suggestion/thought about it will be appreciated... i just "threw the ball" :smile:

cheers,
Alessandro

P.S.: if someone is interested I can upload the test files in .jas, .obj and .blend format (with Lux materials), just drop me a line
 

Attachments

  • AntoineDiamonds_col.jpg
    AntoineDiamonds_col.jpg
    142.2 KB · Views: 1,015
  • AntoineDiamonds_oct.jpg
    AntoineDiamonds_oct.jpg
    143 KB · Views: 1,059
  • AntoineDimonds_Lux.jpg
    AntoineDimonds_Lux.jpg
    142 KB · Views: 984
Last edited:
Archie,

Wow! Now that looks convincing to me!!

Cheers,
Tom

thanks tom,
i'm not 100% happy but the whole thing is extremely dependent on the hdri and while we have no complete file final tweaking makes no sense without the face, pointers and glass.
but the new colorlights should be usable imho.

- archie
 
The diamonds look incredible now.
Going back to the first few posts, I can't believe how far this has come:icon_thumbup:
 
now i used alessandro's hdri and adapted my colorlight mat (can be adjusted in the power and divide nodes)

@alessandro: how is it that octane doesn't utilize your 9600GT ?
the luxrender is very nice btw.

- archie

super cool Archie!
it may not be physically accurate but the look is excellent. maybe I'd try to reduce the "fire" a bit
funny I loaded your colorlight mat in the scene too but then I didn't give it a go :frown:

I'm not able to run Octane with the 9600GT due to an nVidia CUDA driver issue with more than one graphic card on OSX (the on-board 9400m is not totally excluded when I switch to "performance"). I hope the next driver release will solve this problem. (the GPU preview of Lux "SmallLuxGPU", based on OpenCL, let me choose among 9400, 9600, CPU and any combination of them... and it's good for your ATi+i7 too :smile: )

I can't upload the real HDRIs I used in my Cheetah and Octane shots since they're licensed (bundled with other 3d apps)... sorry :(

cheers,
Alessandro
 
hello

Here are my latest tests.

I use the specular setting for a few bright white facets.

Thanks for your opinion.

Antoine
 

Attachments

  • chanel J12.jpg
    chanel J12.jpg
    105.6 KB · Views: 963
  • diams.png
    diams.png
    172 KB · Views: 931
  • Capture d’écran 2010-05-04 à 15.25.56.png
    Capture d’écran 2010-05-04 à 15.25.56.png
    28.7 KB · Views: 922
  • diamdef.c3dmat.zip
    15.6 KB · Views: 441
  • diams2.jpg
    diams2.jpg
    108 KB · Views: 1,194
Back
Top