Last time, the results of simulating smoke were a bit instable and the visual result was a bit bland. Since then, I’ve been irked constantly by spotting smoke everywhere in the – real – world. And sadly, I have to admit I’m a smoker, so there’s no escape from that! :D So back to research! Stam‘s work proved a great inspiration, but his approach needed about 20 iterations over both the density and velocity field. When using BitmapData, you might as well give up right there. Trying something quite different, I decided to combine this approach (I knew I made that blog post for a reason!) on these equations. More or less; I took some “artistic license” (which is a euphimism for not knowing my physics properly :D) . The result is something that looks quite horrible when keeping the smoke source in the same position, but it’s much more interesting when moving it about.

Demo and source (proof of concept style, beware!)

Still not exactly there yet, so the quest continues!