Page 3 of 3 FirstFirst 123
Results 41 to 47 of 47
  1. #41
    Couldn't attach all 4 images. Here is the real-world reference I was referring to.
    Edit: With an attempt at materials 1 and 3. (212 thickness Fe2O3 and 480 thickness TiO2). The green-orange of #4 doesn't seem possible with either of these films so there may be something else going on there.
    Attached Thumbnails Attached Thumbnails Click image for larger version. 

Name:	Colorstream_900x320_tcm2080_101276.jpg 
Views:	7 
Size:	30.7 KB 
ID:	505481  

    Click image for larger version. 

Name:	ColorStream_212-Ti_480.png 
Views:	4 
Size:	524.2 KB 
ID:	505484  

    Last edited by spiderbrigade; 16-Nov-17 at 16:27.



  2. #42
    thanks again for more testing stuff spiderbrigade.
    the thickness units are nanometer (nm).yes i agree ,a gradient is maybe better for comparsion.i have used a texturebump into the thickness input, with very small scale for the big fringles (around 0.05).sandwich render?i guess no,because the transmitting and reflection back calculation (phaseshift) trough all layers is missing then,and thats the main part the thinfilm calculation is for.maybe mixing two material setups ,with two different film layer mats ,could be a good approximation?i havent tryed yet.

    your reference materials looking interesting,do you have more infos about them?

    all these looking like effect coatings for cars ect.i have a idea for a carpaint shader ,but i dont know how to calculating it, or how the code/nodes should looks like.the idea is to have a thin film shader,and the possibily to plugin particles ,with different materials into it.
    but i guess,its not possible with nodes.the calculation maybe yes.but not a plugin system,because for this are programming cases as (if ) ect are needed.at some point cycles is limited for to complicated stuff,that you can programm easy with a single equation codeline for example.

    the colors from your last renderings are going in the same direction,and not far off.at the purple one is maybe the substrate more present?
    Last edited by pixelgrip; 16-Nov-17 at 18:25.



  3. #43
    Yes, effect coatings which you see on cars, plastic goods, nailpolish etc. are what I'm trying to do, hence why I'm including the sparkle/flake texture. Another useful test case would be optical glass coatings which also use metals/oxides. Might work on that later.

    These reference images are from http://www.emd-performance-materials.com/en/coatings/coatings_colorstream/coatings_colorstream.html.

    The diagram I posted is what I've been using as an outline of how these work (what substrate, what film, etc). Unfortunately it's not easy to find exact references to the thicknesses etc. since it's part of their proprietary manufacturing. But if you can nail down the substrate and film parameters, it becomes much more plausible to test different thicknesses to get a good approximation of the effect, at least close enough to feel confident it must be similar​ to the physical product.

    I don't understand the last question, apologies. In those two renders, the orange is a continuation of the Silicon/Fe2O3 series with a thickness of 212. The pink/purple material switches to TiO2 and has a much thicker film. I'm not sure what you mean by "the substrate more present;" these are based on the thin-film only, just split apart into RGB to get the varied IoR for each wavelength.

    Here is another test based on the diagram, in this case Chromaflair which is supposed to be very thin Chromium over MgF2. Reference image claims to be Chromaflair. In this case to get the color bands right I had to "cheat" by varying the layer thickness based on a layer-weight node, with the rationale that the flakes might be presenting a thicker film layer depending on orientation (totally a hack, I know).

    At some point it probably becomes easier to just use a colorramp but I find I get much more interesting emergent interactions with other elements in the scene by trying to get it 'correct.'
    Attached Thumbnails Attached Thumbnails Click image for larger version. 

Name:	ChromaFlair_Cr_MgF2_ref.png 
Views:	9 
Size:	1.32 MB 
ID:	505492  

    Last edited by spiderbrigade; 16-Nov-17 at 18:49.



  4. #44
    edit,your last material looks nice.if you want match the nail material,i would try smaller flakes.

    i guess your pink/purple mat appears a little bit darker because the dark reflections,thats all i think.

    thanks for the link,i have bookmarked some interesting papers about effect coatings a while ago.hope i can find them, i posting this then.

    here it is,happy reading

    http://www.google.com/patents/US7413599

    https://www.google.com/patents/WO200...A1?cl=en&hl=de
    Last edited by pixelgrip; 16-Nov-17 at 19:11.



  5. #45
    flakes try out,gradient as input thickness

    Click image for larger version. 

Name:	sparkle1.jpg 
Views:	13 
Size:	665.5 KB 
ID:	505504



  6. #46
    Originally Posted by JettG_G View Post
    I think you should be plugging the color into a normal glossy shader (same with metal shaders that take n and k values), because if I understand it correctly, thin film calculations are Fresnel equations themselves.

    I have actually been working on a thin film shader that takes both n and k values, too. The equations I've been using follow Pruster's code for thin films in one of the links earlier in this thread. Though, it seems there is a bug in my nodes I haven't been able to solve for the past few weeks (month). I would post the blend file (which also includes the conductor shader group I've created), but I'm still under the 10 post requirement for that, haha (trying to work on that so I can be more involved in the community!). Basically, the node group defines various complex number arithmetic to produce the fresnel (and thin film) equations.

    Multiplying the values (as you show in #35) does not produce correct results; the thin film calculations should already give the correct/close to correct values. You would have to instead adapt the thin film equation you are using to be able to take the k values.

    Good luck!
    for some reason i have overseen this post,sry for that

    yes metal equation into glossy shader is the way to go.
    yes both equations (conductor and thin film)are Fresnel equations.

    i guess you want build a GPU nodes build to? have you a link for your project ,test render?

    multiplying...yes thats the goal ,to implement a pre tested metal equation, to the thin film shader.
    i have build and started this thread for exacly this reason,to be sure the equation gives correct results,and to have some reference datas for comparsion.otherwise how do you know ,your implemented metal code to thin film shader,are gives wrong or right results?

    if you have some reference datas, or something we can use for testing,would be nice

    good luck with your shader



  7. #47
    Originally Posted by pixelgrip View Post
    for some reason i have overseen this post,sry for that

    yes metal equation into glossy shader is the way to go.
    yes both equations (conductor and thin film)are Fresnel equations.

    i guess you want build a GPU nodes build to? have you a link for your project ,test render?

    multiplying...yes thats the goal ,to implement a pre tested metal equation, to the thin film shader.
    i have build and started this thread for exacly this reason,to be sure the equation gives correct results,and to have some reference datas for comparsion.otherwise how do you know ,your implemented metal code to thin film shader,are gives wrong or right results?

    if you have some reference datas, or something we can use for testing,would be nice

    good luck with your shader
    I'm still under the 10 post requirement for posting images and links but...

    If you search up "memo on fresnel" on Google, the first link (by Sébastien Lagarde) has an exact (well, I guess everything we're doing is an approximation) equation for conductor fresnel that you could implement into blender as nodes to compare your other shader to. Surprisingly, that very equation is faster than replacing the math in the dielectric fresnel equation with complex number arithmetic.

    Also, you can find reflectance data for varying angles of incidence (at a specific user defined wavelength) at the the refractive index info website which was mentioned earlier in the thread (I think).

    Oh, I was implying that maybe adapting the thin film equation to perform complex number math might work.



Page 3 of 3 FirstFirst 123

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •