SIGGRAPH goes to #FAccT2026! Congrats to my co-authors @posada.website, @avmenon.bsky.social and @bsky.alexaschor.com
We analyzed every SIGGRAPH, ToG, and SIGGRAPH Asia paper from 1974-2024 to see how race is imagined in graphics research. The results were not good.
arxiv.org/abs/2605.14835
1/
“Skin” is analogized to milk, marble, cheese, wax, and snow. At SIGGRAPH, skin means white skin.
Authors often imply that their models also work on dark skin, but never show it. Post-George Floyd, researchers start grudgingly adding awful-looking Black skin examples to the ends of their papers.
2/
Happy to launch monkscale.com — the home for the Monk Skin Tone Scale. A quick hub for information about the MST, research using the MST, studies about the MST, clinical and tech deployments of the MST, news coverage, and standards adoption across the FDA, ISO, EU AI Act, etc.
#MonkSkinToneScale
I share your concerns... In fact, one of the main features of my scale is its development outside of the CIELAB system.
There's some research on this. The first paper shows the MST outperforms the Fitzpatrick dramatically WRT skin tone classification (0-20% vs 89-92%).
Second paper, I co-authored, shows value of MST for skin cancer detection
www.nature.com/articles/s41...
www.mdpi.com/2313-433X/11...
Excited to share news about Monk Skin Tone Scale - Spatial Frequency Domain Spectroscopy (MST-SFDS). 🧵
MST-SFDS anchors each tone of the MST in chromophore-specific spectroscopy — absorption (μ_a) and reduced scattering (μ_s') coefficients at the wavelength bands where melanin and hemoglobin live.