[Image-SIG] PIL and ICC: littlecms?

Fredrik Lundh fredrik@pythonware.com
Thu, 19 Dec 2002 10:13:39 +0100

kevin wrote:

> Well, I've managed to get things started


> but have a couple problems related to accessing/modifying the image pixel
> data directly in my DLL... I don't think I'm using the right syntax... any help?

I've got no time to dig deeper into this right now, but the following note
discusses the ImagingObject/ImagingNew issues:


(in short, use a Python wrapper).


> is this syntax correct for accessing the pixel data for read/write???

im->image[line] gives you a void pointer to a byte array that contains pixel
data for a line (row) of pixels.  for colour images (im->mode == "RGB"), the
array contains four bytes per pixel:


assuming that the cms layer wants pointers to RGBX data, and that "1" you used
is the number of pixels to process, you could just use:

    for (line = 0; line < im->ysize; line++)
        cmsDoTransform(hTransform, im->image[line], imOut->image[line], im->xsize);

alternatively, to process a pixel at a time, use something like:

    for (line = 0; line < im->ysize; line++) {
        /* pick up line pointers */
        char* in = im->image[line];
        char* out = imOut->image[line];
        for (i = 0; i < im->xsize; i++) {
            cmsDoTransform(hTransform, in, out, 1);
            in += 4; out += 4; /* four bytes per pixel */

hope this helps!