Fun With Dither Algorithms

12.27.13 Marina Galperina

Artist Daniel Temkin launched an online tool that lets user-viewers manipulate “dithering” — the noise that exists at the most deconstructed level of digital imaging. There are several mathematical ways that individual pixels arrange themselves when solving a visual task via algorithms — try out the Dither Studies generator! That’s what I used to make that very jarring GIF above. And I am not very good math! Entrancing.

“It started with me manually trying to expose the dither patterns by giving Photoshop an impossible palette, like giving it a completely green document and telling it to use orange and red to produce it,” Temkin tells ANIMAL. “I was surprised by the complexity of patterns that were exposed, even though the algorithm is so simple.”

(GIF: Prosthetic Knowledge)

“I love seeing the current of pixels in some of the dither pieces; they seem to hold a logic but one that feels irrational,” Temkin ads. “It scans across the pixels and offsets the error to surrounding pixels using the matrix determined by the type of dithering.”

Here’s the algorithmic core — the Floyd-Steinberg (the most popular) vs. the Shiau-Fan dithering whose family of matrices are supposed to “reduce the apparition of artifacts” usually seen with Floyd-Steinberg. They’re trying so hard.


“The main thing about it for me is the irrationality of the algorithm — similar to the conflict set up in my Glitchometry work,” Temkin says, speaking of his Glitchtometry series that translates manipulated sound into beautiful and complex digital landscapes.

Does the generator lend a certain user-viewer collaborative element to Temkin’s practice?

“Sure, it’s a collaboration between the user and the software, which has its own agenda,” Temkin says.

“It’s trying to hide patterns but we make that impossible, so it does the best it can. It’s a very simple system with output that seems insane — builds on work like LeWitt’s [right], which has strict adherence to a system but with irrational output.”

This is the beauty of Temkin’s Dither Studies — like generators that glitch images and text, it employs controls over algorithms and/or code to tame these happenings. It allows a certain degree of authorship in creating effects, a task usually seen as chance events inside the machine.