My feeling is that the problem is the glasing, how to get that feeling the cracks are below the glasing and not on top. I’ll do what I’ll always do, propose a compositing solution, two materials, a ceramic/glase material composited on top of a diffuse render of the textured cracks…
With cycles it is not really possible right now without compositing. You’d need two shaders, the lower one with strong bump mapping for the cracks but the other one (the glazing on top) very smooth. The problem is that as soon as you connect the bump map for the lower shader the top layer gets distorted too. We would need a layered enclosure for that.
I do not agree with bumpmapping, watching ceramic cracking (have similar cracks in some coffey mugs here) there’s no bump, just the thin grey lines - but as I write above, I do agree (for now) compositing two Cycles renders is the way to go.
What you can try, though, is using the solidify modifier, creating an outer shell object with a glass material thus simulating the glasing - but I bet it will be really slow to render and also noisy as the glass-material (especially w/ caustics on) add noise in Cycles…
Edit: Yes, made some quick tests now, it actually works quite well. Make a copy of your object and add a solidify modifier. In wireframe mode, adjust the modifier until you have the secondary object glace thin ‘on top’ of your main object. I then used a diffuse w/ texture maps for the color and a mix shader using glass and transperancy as the glacing. Play around until you get the desired look.
This is a very simplified solution and you can also play around with translucensy & such to get it to look even better, this was just a proof of concept kinda thingie.
“Compositing” is such an “automatic” solution for me now, that I honestly find it difficult to imagine any other approach to such a problem. (Mind you, I’m not saying that that’s a good thing…)
Clearly, it would be very difficult to imagine any sort of “three-dimensional representation” of the cracks in a porcelain glazing. But it does make some tiny amount of sense to visualize the effect as consisting of “what the bowl would look like normally” with some additional “modifier” superimposed (in some way…) upon it. And furthermore, the visual effect really doesn’t have any “3D depth” to it anyhow. The problem is conceptually reduced to two relatively independent problems, each of which has little or no bearing upon the other. “Hence, compositing.”
There are various “visual clues” that give the eye the impression that the cracks are “underneath” the surface … the most important clue being that the specular reflections are unperturbed by the cracks, and the second being that the reflections are somewhat translucent (hence “in front”). But you can convincingly re-create those clues without bothering to monkey with, say, the geometry, and in so doing nonetheless “sell the shot.” And if you can “sell the shot,” hey, mission accomplished: you’re done.
From a production standpoint it’s absolutely the way to go. There’s this misunderstanding about ‘purity’ and other weird concepts, people forgetting it’s all about the final result, NOT how you got there. The best solution IS always the simplest way, everything else is a misconception, hehe… ;D
I look at it this way: What am I going to do in compositing? A simple add, nothing more. If I don’t wan’t to do anything more but add something, why should I spend time setting up a render layer, maintaining an extra shader for that layer and later compositing it if it could be done “in-shader”. Don’t get me wrong, I do a lot of stuff in post but f or this it is too simple of an operation. The simplest way is to do it in-shader IF the engine supports it. But as it currently doesn’t this point is moot. Compared to your solution with additional geometry etc. compositing is the better way to go.
Regarding bump-mapping the lower shader, it does give nice results. Take a look at this for an example: http://www.fundza.com/rman_shaders/subsurf_texture/index.html
When light is moving across the surface it definitely makes a difference. Wherever there are cracks there is a change in surface normals, for a close-up shot I would do it this way.
Because I’m rambling again, TLDR: Don’t create additional geometry, composite it, it would be even better if done in-shader if Cycles would support it, bump mapping the lower surface does make a difference.