Hi Frank,
I would tend to think that wavelets would be a more likely long term prospect for that. But that's just my opinion from having some acquaintance with both through courses at SIGGRAPH.
Personally, I like wavelet compression. (Although, wavelets are a type of fractal, I think there is merit in differentiation. <standing by to get corrected by the mathematicians>) It has a lot of advantages, for example, a partial decoding will give you a partial high level resolving of what you encoded. Among other things, that leads to the ability to use a number of algorithms for operating on the data. Wavelets are an excellent abstraction for geophysical data, for example, because there are things you can do with the wavelet coding that take a lot more time with the raw data.
I think, for example, that one of the big failings of VRML is that it does not specify wavelet encoding for model transmissions. |
