smcameron {l Wrote}:What it looks like in game so far (this is not committed because I'm not yet satisfied with it.)
[awesome pics]
smcameron {l Wrote}:No 3d "thru the window" view yet --- if I do it, it will be a totally software z-buffered flat-shaded triangle renderer that looks like it's from a 1982 silicon graphics workstation. You'd probably rather someone besides me programs that bit, lol. As would I, as would I.
smcameron {l Wrote}:The fluid flows on that planet image were simulated (faked) on 6 square faces of a cube arranged in a field of perlin noise with particles moving to adjacent faces when they leave another face. This works ok, but not great, leaving visible artifacts (though less artifacts than wrapping a seamless rectangle around a sphere). It also does not work well if you want to introduce belts of counter rotating flows parallel to the equator of the planet -- things get a little too tricky for me near the edges of faces and on the top and bottom squares of the cube. I think the better way to do it is do the flows in real 3d on the surface of the sphere, then generate the cube map textures from that data, which is what I'm currently working on, but it's not really there yet.
charlie {l Wrote}:smcameron {l Wrote}:What it looks like in game so far (this is not committed because I'm not yet satisfied with it.)
[awesome pics]
I just looked back and came across this in the 2nd post of the thread:smcameron {l Wrote}:No 3d "thru the window" view yet --- if I do it, it will be a totally software z-buffered flat-shaded triangle renderer that looks like it's from a 1982 silicon graphics workstation. You'd probably rather someone besides me programs that bit, lol. As would I, as would I.
Hrm... are these two the same guy? I'm unsure.
NaN {l Wrote}:Beautiful idea to use curl of noise function to get divergence free velocity field.smcameron {l Wrote}:The fluid flows on that planet image were simulated (faked) on 6 square faces of a cube arranged in a field of perlin noise with particles moving to adjacent faces when they leave another face. This works ok, but not great, leaving visible artifacts (though less artifacts than wrapping a seamless rectangle around a sphere). It also does not work well if you want to introduce belts of counter rotating flows parallel to the equator of the planet -- things get a little too tricky for me near the edges of faces and on the top and bottom squares of the cube. I think the better way to do it is do the flows in real 3d on the surface of the sphere, then generate the cube map textures from that data, which is what I'm currently working on, but it's not really there yet.
You say you have tried Latitude / Longitude Projection (seamless rectangle)? Did you scale vx or dx to account for changing circumference cos(lat)? Should be simpler than trying to get the Cubed-Sphere Grid (cube map) right I think.
Edit:
Using spherical coordinates for particles positions and velocity should work for cubemaps I think. You just have to figure out the transform from spherical to cubemap coords
Users browsing this forum: No registered users and 1 guest