jcupitt
commented
9 years ago I had a quick look at the normalise function, it looks nice.
You could make it a bit more concise with vips_object_local_array(), I expect you know. For example (untested):
VipsImage **t = (VipsImage **) vips_object_local_array(hook, 7);
// normalize the luminance band in LAB space:
if (vips_colourspace(image, &t[0], VIPS_INTERPRETATION_LAB, NULL) ||
vips_extract_band(t[0], &t[1], 0, "n", 1, NULL) ||
vips_extract_band(t[0], &t[2], 1, "n", 2, NULL) ||
vips_stats(t[1], &t[3], NULL)) {
return Error();
}
double min = *VIPS_MATRIX(stats, 0, 0);
double max = *VIPS_MATRIX(stats, 1, 0);
if (min == max) {
// Range of zero: create black image
if (vips_black(&t[4], image->Xsize, image->Ysize,
"bands", 1, NULL )) {
return Error();
}
} else {
double f = 100.0 / (max - min);
double a = -(min * f);
if (vips_linear1(t[1], &t[5], f, a, NULL) ||
vips_bandjoin2(t[5], t[2], &t[6], NULL) ||
vips_colourspace(t[6], &t[4], image->Type, NULL)) {
return Error();
}
}
VipsImage *normalized = t[4];You could also consider taking a histogram of L* and setting the range to 1% and 99%, ie. letting the top 1% and the bottom 1% over- and under-expose. This would make it less sensitive to noise and probably produce a more pleasing image in more cases.
vips_percent() does almost this, you'd swap your vips_stats() for this code:
https://github.com/jcupitt/libvips/blob/master/libvips/histogram/percent.c#L82
lovell
blowsie
I'd like to be able to use sharp on the Windows platform. Because
libvipsis available for Windows, I assume that lack of support is mostly in figuring out the build script.Has this been investigated already? If so, what approaches have already been tried and roadblocks encountered? Knowing what's been tried so far, if anything, will help me or other potential developers come to a working solution quicker.