I now have realtime fractals within something like 5 ms CPU time per frame. Better than in my dreams! I must confess I needed much more computations than I expected. Alas, I’m not the hero on this work. Nowadays CPU power are simply stunning!!!
Pleasant consequence: in this speed quest, I have gained a full set of time benchmark methods, lol.
I still got bugs, but landscape looks more and more realistic to me. In fact, that’s not the real problem anymore. Fractal or not, I won’t be able to store something like +13 million vertex per planet anywhere… I need real time processing.
“Real time”… kind term for meaning less than 17 milliseconds of CPU time per frame (on a 60 FPS basis). Ogre has a very impressive speed on its side, nothing to blame. Thus, optimization is needed on my own formulas. I use a lot of cubic interpolation. Maybe could I use GPU i guess? That sounds unneccessarily complicated… I might be able to fit within the CPU. Hope so!