Hao Xie

Here comes the Sun 🌞 …when the Helmholtz-Kohlrausch (H-K) effect is incorporated. A thread 🧵1/n

Using biophysically generative process (analogous to Asano’s individual CMFs), a comprehensive & physically realistic modeling, reconstructed from RGB capture, is proposed & holds huge potential for spectral rendering, imaging simulation & skin perception. Plz check out soon. 3/3

Skin spectral data such as those from Leeds-Liverpool and Stanford ISET have been useful but w. some limitations in terms of spatial characterizations, diversity, etc. 2/3

Delighted and a bit unrealistic to be on the author list of my first graphics paper, <A Hyperspectral Space of Biophysical Skin Tones for Inverse Rendering of Skin Properties>🟫

@troy_s If the output is XYZ, that could be preserving the white point for equal-energy spectra.

@HA0X1E It only applies to luminance I believe, and concatenation will not change this?

I did a spit take the other day when I heard that all colourimetric matrices must sum to unity across rows. This is as best as I understand things, false. Only the luminance Y position of the classic XYZ matrix sums to unity, representing 100%. Remember where I said the base +

The full program of @ImagingOrg #CIC30 is now online & registration is open (at a discount until Oct 16) tinyurl.com/2hca79wb The best part of the conference is meeting the #color and #imaging crowd from around the world. A few photos from past events to whet your appetite :)

@yzhu88 The mixing needed more stirring, but the part where the mixing happened was amazingly reasonable (the green).

@HA0X1E Great idea. I thought about something like "mixing two colors" or "mixing two pigments in a paint", but haven't got an account yet ;)

I asked DALL·E 2 to imagine the blackest black, the whitest white, etc. to see its "gamut". Wondering if the result is of any possible use for wording the instructions in (human) psychophysics?

@thecolorsapp cc. @HellwigLuke

"...large color differences saturate relative to the Riemannian prediction." Nice commentary by David Brainard doi.org/10.1073/pnas.2… Seems the idea above is consistent with the law of diminishing returns for both color and money.

It looks fun to monetize color in an experiment (with fast & "accurate" decisions). As we're thinking about *large* color differences, I wonder whether the opposite (to colorize money) can help...💵💶💷

To summarize, the H-K effect is not necessarily an undesired artifact to be compensated (most times it should). Luminance & brightness can be reconciled as they serve different purposes for both artists & scientists. See also doi.org/10.1364/JOSAA.… & doi.org/10.1364/JOSAA.… n/n

The vessels in the BG become less salient in my conversion (probably they should be). Color contrast needs non-trivial care when doing the 3D-to-1D mapping. See the benchmarking results where sometimes contrast polarity did not even need to be correct: cadik.posvete.cz/color_to_gray_… 6/n

Here comes the Sun 🌞 …when the Helmholtz-Kohlrausch (H-K) effect is incorporated. A thread 🧵1/n

This Wednesday, I will discuss the Luther condition for individual observers at aic2022.org. It was a long-hatched idea and a fun side-project.

It was hugely inspirational. I also love the stories that the prism from his mother sparked his interest in vision at 5 yo, and his kids' jellycat motivated him to switch to tactile sensors. And the remarkable checker shadow illusion was a result of two-year iterative innovation.