Legacy ImageMagick Discussions Archive

Hello, I use on this image:

This code: This is meant to result in that it gets quantized to 15 colors, using no dither, and using XYZ colorspace as the basis for quantizing.
Here is the result:

However, you can count that there are only 11 unique colors in this image. Did I do something wrong? Because if not, this seems like a bug since this image had 22 colors, so it should have been possible to reduce it to 15 colors.

-colors will quantize to at most the number of colors you request. But it may be less. See https://imagemagick.org/script/command- ... php#colors

fmw42 wrote: ↑2019-02-10T23:32:15-07:00 -colors will quantize to at most the number of colors you request. But it may be less. See https://imagemagick.org/script/command- ... php#colors

Ah, thank you. I guess a mod can move this to some better forum since this isn't a bug. Is there a way to get it to use that exact amount of colors for maximum detail within the quantized set of colors? I tried adding -treedepth 2 or -treedepth 8 to the command, but the result is not better.

P-Tux7 wrote:Is there a way to get it to use that exact amount of colors for maximum detail within the quantized set of colors?

No.

You can script a loop to vary N in "-colors N", and find the actual number of unique colours. For example, "-colors 21" will return 13 unique colours, but "-colors 22" returns 22 colours.

You can try a kmeans type solution with seed colors. I have one for bash unix shell script at my line below and user snibgo has one for Windows at his web site shown below his post.