I was just looking into why error diffusion was occurring in my images when I went from RGB to P with a 140-color image. Unexpected. I see in the code that Floyd-Steinberg error diffusion will be applied on conversions from "RGB" to "P", regardless of whether it's needed, unless the optional "dither" parameter is specified. So I changed my code to specify "dither=Image.NONE" where appropriate.
Would it make sense to modify that default behavior to dither only if needed? That is, when converting from "RGB" to "P", only dither if the number of colors is greater than 256? Bill _______________________________________________ Image-SIG maillist - Image-SIG@python.org http://mail.python.org/mailman/listinfo/image-sig
