roland p, on Mon Jan 16, 2012 3:15 AM, said:
Thomas Jentzsch, on Mon Jan 16, 2012 2:56 AM, said:
Hm, I wonder if a "real" 3D Asteroids would work. Like a FPS, just with Asteroids and UFOs attacking from all directions.
Does such a game exist?
Maybe alianator comes close? : http://www.lemonamig...ils.php?id=1609
That's a sort of 3d-defender.
Thanks, I will check it.
Btw. the demo you posted looks really good. How many divisions/multiplies does the code? It looks damn fast for a vcs...
Quite a lot.
All multiplications (IIRC there are no divisions) are done using square-tables, so they are pretty fast. How detailed do you want to have it? I have to transform all points and have to check the visibility of each polygon.
Note: For the polygon visibility I once switched to using a predefined normal vector, but I found out that this wouldn't work in a real 3D world. So I would have to create the normal vector from 3 transformed points of the polygon. Which means a few optimizing steps wouldn't work there anymore.
Also the math precision is currently very limited, so here it would become more complex too. And the step which transitions a whole object in space is missing too.
Finally all lines are assumed to be completely visible now. So no clipping and only backface culling
So for a real 3D game, this part would for sure become significantly slower and I would have to find some time somewhere else e.g. less complicated bankswitching (I need two double buffers of 1k each, but here I have to use one 1K buffer and four 256b buffers), co-processor etc.
Edited by Thomas Jentzsch, Mon Jan 16, 2012 6:48 AM.