raveling@isi.edu (Paul Raveling) (02/01/90)
Many of you recall that I'd expected to add a new color quantization algorithm to the Img Software Set by about now... Well, since early January I've had about 10 minutes to work on it. It'll probably be available before the next ice age arrives. Unfortunately this is an after-hours project that's not what ISI pays me to do, so the necessary detailing may be slow. BTW, a plus from that 10 minutes of work is that I tried out a different distance function that looks more promising, involving n*log(kn) terms. A minus is that I realized that some colors could be assigned to the wrong parent node in the tree in the last version. A quick crude fix DOES get prime image quality, but it costs speed. The next 20 minutes or more of work will involve a couple algorithm changes (not major) to keep best quality and regain speed. But that 20 minutes could stretch... I'm still thinking "Februaryish" for having code that's fit for public consumption, despite having a week off for skiing. ---------------- Paul Raveling Raveling@isi.edu