Thanks for the info, Matthew. In looking at the docs again, I see now something that I should have noticed earlier, that the arguments to set-argb-pixels include a bytes?, and that should have been sufficient for me to answer my own question. Sorry for the noise. <div class="gmail_quote">On Thu, Sep 13, 2012 at 9:07 AM, Matthew Flatt <<a href="mailto:mflatt@cs.utah.edu" target="_blank">mflatt@cs.utah.edu</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> <div class="im">At Mon, 10 Sep 2012 13:00:55 -0700, Danny Yoo wrote: > Hmm. Something VERY strange is happening with the alphas being > produced by ico->argb vs the alphas expected by the bitmap functions. > They seem... flipped! </div>The problem seems to be in `argb->bitmap' from `mrlib/cache-image-snip'. That function creates and old style bitmap+mask combination (insteda of a bitmap with an alpha channel), and it uses "alpha" values as mask values (instead of inverting an alpha value to arrive at a mask). Here's a version of `ico->bitmap' that uses `racket/draw' directly and creates a bitmap with an alpha channel: (define (ico->bitmap ico) (define w (ico-width ico)) (define h (ico-height ico)) (define bm (make-bitmap w h)) (send bm set-argb-pixels 0 0 w h (ico->argb ico)) bm) ____________________ Racket Users list: <a href="http://lists.racket-lang.org/users" target="_blank">http://lists.racket-lang.org/users</a> </blockquote></div>