The basic explanation is Newton's scale of interference colours: the interaction of surface tension and gravity means the film is thinner at the top than the bottom, so you get a range of colours that are a function of film thickness.
If the whole setup were perfect and static, you'd just get a series of colour bands. However, turbulence and rising bubbles produce very interesting patterns due to the thickness variation they cause in the film. As I mentioned previously, Andrew Davidhazy's Dynamic interference colors in thin soap films gives more specific tips - but this brief trial, with no setup whatsoever, suggests it's not going to be wildly difficult to get good results.
- Ray