Time to update the blog, since my 3D noobish-to-nerdish research was silent for a while and yet, luckily, it has gone further than I could expect (at least in such a short time).
In the last few weeks, I took some time to experiment with fluids, particles, nDynamics, plus fur and hairs of any sort.
Also, I spent some time and energies studying PRMan and 3Delight. Dealing with that, I’d add my obvious voice to the choir in confirming that Advanced RenderMan is just the way to go, for anybody approching the renderer who’s still wondering about the best resource to begin with. That’s unbelievable how manuals, which are well focused on theory and method rather then merely and mechanically describing the API, can be valuable even after more than 10 years.
That said, my first tiny update is about fluids to simulate smoke. My PRMAN knowledge doesn’t cover volumetric effects, yet, then I just chose to go classic, and used mental ray for my very first test.
Here it is:
Not too much is to be said about it: I just followed a couple of tutorials, and messed up with attributes and textures. I didn’t play around with lights too much, since I was looking for a neutral result, but I’ll produce some more characterized rendering when I’ll find the right shot to composite it.
Greatest lessons I learnt:
- never test the fluid at resolutions lower than 80 or so, even when the scene setup is in the earliest stage, and above all when fields are influencing the simulation: the results are so far from the ones coming out of higher resolutions, that they’re just not indicative at all
- whenever a complex behaviour is the main aim for the simulation, the emitter’s Density/Voxel/Sec attribute is to be set to high values: 50 to 100 is generally a good starting point, to me
- with my machine’s 4Gb of RAM, higher resolutions than 230 are simply prohibitive, and caching fluids will always fail computation on many frames; of course that depends on the grid size, but I observed that 230 is a good resolution average for fluids whose containers tend to expand a lot, during the simulation