21 Replies Latest reply: Aug 14, 2012 8:18 AM by conroy RSS

    caustic donut

    conroy Community Member

      CS6 Extended raytraced caustics.

       

      caustic-donut.png

        • 1. Re: caustic donut
          Noel Carboni Community Member

          Cool shape, but what does the word "caustic" imply?  My train of thought got derailed.  It happens. 

           

          -Noel

          • 3. Re: caustic donut
            Noel Carboni Community Member

            Thanks.

             

            Second question:  You're showing these images with what look like "light panels" in them...  I've been looking for ways to do that and have so far not stumbled upon how...  I've been looking at the various options for lights, and I've rotated my views so the light sources should be visible, but nothing's showing up.  Perhaps I've missed something obvious...  Can you help "shed some light" on how to get those in the image, and so that the nice reflections are generated? 

             

            -Noel

            • 4. Re: caustic donut
              conroy Community Member

              I'll illuminate you, Noel.

               

              High dynamic range panoramic images for image-based lighting. Adobe provides some as optional 3D content:

               

              http://www.photoshop.com/products/photoshop/3d

               

              To emphasise the caustics in the image above, I put the HDR in the donut material's environment slot only. If I had used it as global environmental IBL, it would have directly illuminated the square, too, which obviously would have reduced the contrast of the indirect illumination from the "chromed" donut.

              • 5. Re: caustic donut
                Noel Carboni Community Member

                Oh, snap!  I was trying to create everything from scratch.  It didn't occur to me to download some presets.  Thanks!

                 

                I was doing okay making innertubes though... 

                 

                Tube.jpg

                 

                -Noel

                • 6. Re: caustic donut
                  conroy Community Member

                  Regarding IBL, see this thread:

                  CS6 3D file size enormous bloat - where's the compression?

                   

                  It appears that PSD and PSB files do not support compression of 32-bit float images!


                  Images which can be losslessly compressed to a few hundred KB can gobble up hundreds of MB in a Photoshop document.

                   



                  • 7. Re: caustic donut
                    Noel Carboni Community Member

                    This 3D stuff is great fun.  I take it you've some experience with it.  I only "graduated" from 2D photos with Photoshop CS6, when I moved up to Extended.

                     

                    InnerTube.jpg

                     

                    -Noel

                    • 8. Re: caustic donut
                      conroy Community Member

                      I've been a geek who enjoyed coding Renderman shaders many moons ago. There's a pun in there. I wonder if any of the regulars here will get it.

                      • 9. Re: caustic donut
                        Noel Carboni Community Member

                        "Geek", "Renderman", "moons", et. al. don't ring a bell with me, but since you're a Mac user we may not have crossed paths much before the forums were smashed together (thinking "large hadron collider").

                         

                        All I can say is thanks for the tips, from a fellow geek of different feathers.

                         

                        InnerTubeDeluxe.jpg

                         

                        -Noel

                        • 10. Re: caustic donut
                          Noel Carboni Community Member

                          And of course we have the frosted glass paperweight on a glass desk, and a nice shiny black innertube bouncing in the air...

                           

                          FrostedGlassToroid.jpg

                           

                          ShinyInnerTube.jpg

                           

                          -Noel

                          • 11. Re: caustic donut
                            conroy Community Member

                            There are a couple of astonishing weaknesses in the CS6 raytracing material which prevents realistic rendering of common opaque and translucent dielectric materials such as ceramics, glasses, plastics, gloss paints and lacquers, and water:

                             

                            1. There is no modelling of the Fresnel effect. If the user specifies a refraction index other than 1.0 (regardless of opacity) then it should be assumed that a dielectric material is being simulated and the Fresnel effect should be modelled. Without the Fresnel effect, reflection intensity is constant at all angles of incidence to a surface, which is a property of metals. With a dielectric, e.g. glass, reflections are progressively weaker as the angle of incidence approaches being face on to the surface. The fraction of light that isn't reflected is transmitted into the surface. The CS6 material fails to implement that simplistic model which can be found in every other raytracer.

                             

                            2. Translucency is bizarrely modelled by the CS6 material. Keeping to the basics and ignoring Beer's law (although its inclusion will greatly increase realism), translucency should be a multiplication of transmitted light colour by the material's filtering colour (the one in the "Diffuse colour" parameter being used for example). The CS6 material doesn't produce that effect and, instead, the filtering factor is strangely decreasing as the light intensity increases. I wouldn't be surprised if some relationship to the opacity parameter is responsible for this strangeness. An opacity parameter makes more sense as a "ghosting" control for the entire visibility of a material.

                             

                            I said these weaknesses are "astonishing" because refraction is modelled by the material and the renderer is otherwise capable of good results and supports image based lighting and indirect illumination.

                             

                            Will this be corrected before CS7, I wonder.

                            • 12. Re: caustic donut
                              Noel Carboni Community Member

                              I don't have the depth of knowledge you do, but I get the impression that all the "magic" is occurring on the faces, and the model doesn't really have an "inner material" per se - kind of like a clear glass with a colored coating, or maybe hollow plastic with colored walls.  So it's not a matter of the ray tracer doing it wrong so much as an incomplete modeling of reality.  Thus it's impossible to apply Beer's law per a given material if there's no sense of a material through which the light is passing.  Does that seem to make sense?  It's purely an observation.

                               

                              Plus I'm finding a few things that may be ray tracing glitches...  This model of an extruded R, for example, seems to be missing some color in a strange way in the spaces where it's cut through.  There only appear to be a few places where this is occurring, and note that things look okay on the right side of the R.  Notably I didn't let the rendering finish, but it doesn't seem likely to change something as radical as that - everything seems to be just refinement.

                               

                              RColorProblems.jpg

                               

                              -Noel

                              • 13. Re: caustic donut
                                Noel Carboni Community Member

                                More on that...  Note the coloration of the OpenGL draft approximation of same...

                                 

                                OpenGLDraft.jpg

                                 

                                -Noel

                                • 14. Re: caustic donut
                                  conroy Community Member

                                  I don't have the depth of knowledge you do, but I get the impression that all the "magic" is occurring on the faces, and the model doesn't really have an "inner material" per se - kind of like a clear glass with a colored coating, or maybe hollow plastic with colored walls.  So it's not a matter of the ray tracer doing it wrong so much as an incomplete modeling of reality.  Thus it's impossible to apply Beer's law per a given material if there's no sense of a material through which the light is passing.  Does that seem to make sense?  It's purely an observation.

                                   

                                   

                                  Don't get hung up on Beer's law, which is very often not coded into materials/shaders - it's icing on the cake. I wish I hadn't mentioned it now. I'll come back to it in a moment, though.

                                   

                                  The raytracing engine itself is producing nice results, as I said, but the material modelling is bizarrely weak (not because Beer's law is absent) because a staple of raytracing is missing, namely the Fresnel effect, and translucency is just plain weird. Fresnel is essential for basic "photorealistic" rendering of dielectrics, e.g. glass, and its omission is a pity.

                                   

                                  You are correct that the raytracer deals with surfaces, but that doesn't preclude the modelling of material in the volume bounded by these surfaces. We're back at Beer's law now. It can be simplistically represented by a couple of lines of code where the "sense of a material through which the light is passing" is just the length of the ray that's being traced inside an object.

                                  • 15. Re: caustic donut
                                    Noel Carboni Community Member

                                    I think we're in agreement, though I don't know the proper terms to express things in.

                                     

                                    I've found that Photoshop's ability to combine the 3D results with other layers is kind of cool.  It leverages my knowledge of how to use Photoshop, and as you probably saw from the other thread on the gun logos I have been able to do a passable job of making a backlit glass effect.  It doesn't stand up perfectly to close scrutiny, but it looks pretty good - so again we're in agreement - the raytracing engine has produced nice results, and Photoshop has been able to use them well in a bigger design.

                                     

                                    -Noel

                                    • 16. Re: caustic donut
                                      Noel Carboni Community Member

                                      By the way, one thing I found impressive:  I fooled with Photoshop in one session doing experimentation and rendering after rendering for something like 8 hours straight yesterday afternoon, without a single Photoshop fault.  It's not like Rendering was fast - even with 8 cores on the job - but it didn't run into any brick walls, which was nice.  ATI Catalyst 12.6 seems to be pretty good for Photoshop.

                                       

                                      -Noel

                                      • 17. Re: caustic donut
                                        conroy Community Member

                                        Here's my Photoshop render from the Glock thread and the same red glass object (a box with a square tunnel through it) rendered by another app. In the latter image, see Beer's law in action - the longer a light ray's path through the glass, the more it is absorbed, producing a deeper red.

                                         

                                         

                                        redglass-Ps.png

                                         

                                         

                                        redglass-other.png

                                        • 18. Re: caustic donut
                                          Noel Carboni Community Member

                                          Plus apparently you're able to make it completely clear but at the same time deeply colored, which was my original complaint.  Opacity isn't the same as coloration.  I remember my Chemistry teacher in college, one Dr. Baxter, a lifetime ago flunking people for not differentiating between "clear liquid" and "colorless liquid".  Fortunately I got the distinction early on and aced his course.

                                           

                                          -Noel

                                          • 19. Re: caustic donut
                                            conroy Community Member

                                            It is very strange that Adobe have implemented such a poor material model when the raytracing engine is quite certainly capable of so much more realism. I bet Adobe's solution will be a $200 upgrade to CS7.

                                            • 20. Re: caustic donut
                                              PECourtejoie ACP

                                              Conroy, did you post at the feedback site regarding the remarks about the materials, with examples?

                                              • 21. Re: caustic donut
                                                conroy Community Member

                                                No, Pierre. If I remember correctly, Chris Cox previously stated in this forum that there is no need to report problems both here and at the feedback forum.

                                                 

                                                Unfortunately, when no acknowledgement by Adobe is made of a reported problem, there is no way to know whether the report has been missed or not. I have been instructed by Chris that I am not to bump unacknowledged threads in order to discover whether a report has indeed been noted or missed. I will assume an absence of acknowledgement means that a reported problem has not been noticed or the problem is of no interest to Adobe.