I don't know why uncompressed is even an option. Always compress your raw data, unless you have a good reason not to (I'm not even sure there ever is one anymore, for normal users anyway) - lossless means "no loss" - none... original is reconstituted bit-for-bit when decompressed...
Maybe if you were decompressing on a computer which didn't have a math processor or something, you might want them uncompressed, but modern CPUs can decompress faster than they can read from disk, so lossy compression is:
* faster
* smaller
i.e. there is NO down side.
I'd love for anybody to point out real scenarios where that is wrong.
Rob