Hi Not sure of optical questions should go here to general physics questions, but here goes. I have a doubt, basically, what consists of a reflection? If i have a perfectly smooth surface, then the specular highlights are infinitely thin, and you should have perfectly mirror-like reflections, a mirror like surface. If you have a rough surface, rough as in microstructure (say, like the V groove microfacet models in the Cook-Torrance BRDF), then we're modelling the probability of these microfacets being oriented in such a way that they reflect light towards the viewer, with a distribution function, say, the Beckmann probability distibution function. You then calculate the shadowing and obscuration effect these microfacets would have on the adjacent microfacets, and finally calculate the amount of reflected light from these microfacets using a Fresnel term (dependent on the incident (light) direction, and the microfacet normal (since we want the amount reflected off the microstructure, not the macrostructure). So far so good (please correct me if i'm wrong) ? In this latest case, of a rough surface, then specular rays are reflected in a series of directions by the microstructure, this would mean that a reflection would also be scattered in these directions, so, you would have a fuzzy/blurry reflection. Is this correct? My doubt is, is there any difference at all from a reflection ray or a specular ray? From a point of view of shader writing for instance. I'm thinking that perhaps perfectly smooth surfaces would have infinitely small specular highlights, and coherent, mirror like reflections, while rough surfaces would have scattered specular highlights, and incoherent, fuzzy/blurry reflections. If this is the case, then from a physics point of view, there is no distintion between the process of forming a specular highlight, and the process of forming a reflected image. The only difference would lie in whatever way you decided to implement this in a shader, is this correct? What about global illumination? In RenderMan renderers such as 3delight, pixie, etc... you can gather rays using 2 prebuilt distributions, cosine or uniform. Say you want to gather reflection rays along a direction vector V, with a solid angle X radians, to build a reflection image, and you use an anisotropic distribution, or an isotropic distribution (say, the Ward distribution for instance), would this be equivalent to building a specular term? To creating the contribution of the specular highlights, since a narrower sampling cone would mean a smoother surface, therefore a sharper reflection, and a wider sampling cone would mean a rougher surface, therefore a fuzzier/blurrier reflection? What's happening in the physical/optical point of view in the process of forming a specular highlight and a reflection, are they different processes at all, or exactly the same, and the distintion is artificially created by whatever coding/shader/rendering implementation you might use? Thanks in advance for your time. Best regards P.S.: in the microfacets case, what happens to light when the spacing between microfacets is equal to a particular light wavelenght, say 550nm? Is there a shift towards a lower or higher wavelenght? Or just a cancelation? Sancho R.