I find writing an RSR way easier than reading (as there I got some problems ;-). Encoding an RSR for Poser is really easy: as the number of pixels is fixed in the header you just write out pixel after pixel line by line, so 3 or 4 bytes (if with alpha) of color information (that's what the 3/4 means in the header ;-)). There's a repeat count, too, should be in front of the color information... That's just from memory though, but should be all you need. Try kawecki's PNG to RSR converter on a few images or ask him and you'll easily find the algorithm to write out the RSRs.