The main problem i got is the performance, especially on arm5 and 1core cpu devices.
For arm5 i simulated float-point arithmetics as int and for 1core cpus implemented complex optimizations of standart marching cubes alhorithm. But i am still not sure it works fast enough on low-performance devices.
These optimizations include linear memory usage, adjacent edge calculations for storing already calculated points and half grid size calculations according z axes direction.
If you have android phone or some suggestions about any more optimizations, please help me improve my wallpaper. Thanks.
P.S. can be downloaded here
Edited by Hamster Beat, 14 June 2012 - 04:23 PM.