This is a 16 color palette reduced image using my own custom palette generator algorithm. The palette it detected and in use is in the top right.
Based on this post right here…
This image has this lovely crustiness to the limited color palette that really speaks to my days using fishbowl Macs in the computer lab.
Yes indeed 👍
But there’s a little something different I did here, I made my own custom palette reduction algorithm, that finds the best palette for a given image.
Yeah it’s not a new idea by far, but my algorithm seeks the edges of the color space first, yielding higher color saturation than most other palette reduction techniques.
I’m really impressed with how well it retained the fuzziness of their faces and on the traces on the upper one’s wing segments.
I didn’t exactly reinvent the error diffusion dithering algorithm, I just reinvented palette reduction tailored to a specific image, that gives the most optimal palette for the given image.
I wrote the algorithm like 18 years ago, kinda just testing myself to see if I even remember how it works in its incomplete state. Apparently so… 👍
Great photo and great example of colour efficiency.
Makes me think of the recent stand up maths video on dithering.
He’s doing ordered dithering plus random generated noise there. That’s a bit different than error diffusion dithering, the noise comes from the image itself via a feedback loop.



