利用者:Sebbas/GSoC 2016/Reports/Week 1

提供: wiki
移動先: 案内検索

Week 1

What's been done

  • Some cleanup in my 'fluid-mantaflow' branch. Merged master and removed deprecated smoke code. Some Scons files where still there too. (rBc65d58a3846d, rB7a8fd7068ef7)
  • It turns out that my prototype did not scale very well: Loading the fluid objects in the form of .obj into Mantaflow can become ugly (you'd have that .obj file somewhere on disk) and might underperform. Especially when dealing with large meshes. I discussed this with my mentor and he advised me to just fill up a levelset grid which indicates what's inside and what's outside the mesh (e.g. -1 in and 1 outside).
  • So I tinkered with my best friend smoke.c and used the BVH trees to calculate such a map. To load it into Mantaflow I used the same structures I am using with all the smoke & fire grids. (rB7872e2fc3b04)

Next up

  • Discuss my approach to create a "map of points indicating points inside and outside a mesh" on IRC. My implementation works but I feel there might already exist a function or so for that (which I'm just not aware of).
  • Clean up smoke.c so that smoke & fire work again. I had to disable a lot for liquids to work. (rB240ed366ce70, rB21d382fdfe84)

Questions

  • How problematic is it to "kill" a physics modifier (backwards compatibility, issues in .blend files)? I am thinking of totally removing the "Fluid" modifier and repurposing the smoke code to work for liquids as well.