Huffman encoding is the most common form used in JPEG for the lossless coding of the discrete cosine transform (DCT) coefficients. There is also arithmetic encoding which can produce smaller files. But the big savings in size comes from reducing number of bits for the coefficients of the DCT, and this is controlled by the quality slider. Different JPEG compressors have different meaning for 90% etc (IOW something like, say 85% isn't the same in Photo Mechanic as it is in another program with 85% quality). In PM, 100% is essentially a lossless encoding (other than an occasional rounding error of 1 level in the DCT math) and can cause file sizes to be quite large, especially with a noisy source that has been cropped on a non 8x8 boundary because the compressor is trying to accurately represent all the noise and artifacts.
PM uses RGB photometric interpretation.
--dennis
