This message was deleted.
# helps
Slackbot
04/26/2023, 1:46 PMThis message was deleted.
m
Max Bo
04/26/2023, 1:47 PMI am trying to see how different color schemes behave under different color vision deficiencies. This is what I've plotted so far:
Max Bo
04/26/2023, 1:47 PMproduced with the following code:
Copy code
{
const none = (t) => culori.formatHex(t);
const greyscale = (t) => culori.formatHex(culori.filterGrayscale()(t));
const deuteranopia = (t) =>
culori.formatHex(culori.filterDeficiencyDeuter()(t));
const tritanopia = (t) => culori.formatHex(culori.filterDeficiencyTrit()(t));
const protanopia = (t) => culori.formatHex(culori.filterDeficiencyProt()(t));
const turbo = (t) => d3.interpolateTurbo(t);
const viridis = (t) => d3.interpolateViridis(t);
const plasma = (t) => d3.interpolatePlasma(t);
return Plot.plot({
color: { type: "identity" },
nice: true,
aspectRatio: 322 / 420,
fx: { tickFormat: (f) => f?.name },
fy: { tickFormat: (f) => f?.name },
marks: [
Plot.raster({
fill: (x, y, { fx, fy }) => {
console.log(fx, fy);
if (fx && fy) {
return fx(
fy(monaLisaLightness[Math.floor(322 - y)][Math.floor(x)])
);
}
},
fx: [none, greyscale, deuteranopia, tritanopia, protanopia],
fy: [turbo, viridis, plasma],
x1: 0,
y1: 0,
x2: 420,
y2: 322
})
]
});
}Max Bo
04/26/2023, 1:49 PMyou'll notice a few things:
1. some of the faceting works, and the functions are getting passed down just fine
Max Bo
04/26/2023, 1:50 PM2. you'll see that
console.logfillfxfyundefinedMax Bo
04/26/2023, 1:52 PMin fact, the squares that nothing was rendered in corresponds to when
fxfyundefinedf
Fabian Iwand
04/26/2023, 1:56 PM(I don't have an answer, I just wanna say thanks for sticking to this 🧵 format! 🙏)
😁 1
➕ 1
🙏 1
Fabian Iwand
04/26/2023, 2:00 PMCan you share this as a notebook?
m
Max Bo
04/26/2023, 2:00 PMsure
Max Bo
04/26/2023, 2:01 PMf
Fabian Iwand
04/26/2023, 2:03 PMMy first instinct would be to plot one of the broken combos in a separate cell to see the produced values
m
Max Bo
04/26/2023, 2:04 PMyep, works
Max Bo
04/26/2023, 2:04 PMusing
Copy code
fx: [protanopia],
fy: [turbo],f
Fabian Iwand
04/26/2023, 2:06 PMsomething about the domain perhaps?
m
Max Bo
04/26/2023, 2:09 PMmaaaaybe? I dunno what could be wrong with the domain configuration given that its working with some of the images 😬
f
Fabian Iwand
04/26/2023, 2:10 PMwhat's also noticeable is that every row and every column only has one entry, and that repeating the functions in the domain doesn't work either
m
Max Bo
04/26/2023, 2:12 PMokay I made a subtle error with the raster bounding but fixing that didn't affect the bug
f
Fabian Iwand
04/26/2023, 2:12 PMyeah, tbh I have no idea what's wrong
m
Max Bo
04/26/2023, 2:12 PMI think it may be time for me to look at the source code because I'm worried this is a bug lol
f
Fabian Iwand
04/26/2023, 2:14 PMI think I'd try to build back to only pass through the initial values, ignoring the actual domain
Fabian Iwand
04/26/2023, 2:39 PMI give up. I tried to make sense of the faceting based on https://observablehq.com/@observablehq/plot-raster?collection=@observablehq/plot#cell-762 (note how each domain has 4 values?), but I don't see a clear pattern.
Maybe @Fil can help?
Fabian Iwand
04/26/2023, 2:49 PMOK, I've grokked it. You need to supply domain values for every facet:
Copy code
fx: [
none, greyscale, deuteranopia, tritanopia, protanopia,
none, greyscale, deuteranopia, tritanopia, protanopia,
none, greyscale, deuteranopia, tritanopia, protanopia,
],
fy: [
turbo, viridis, plasma,
turbo, viridis, plasma,
turbo, viridis, plasma,
turbo, viridis, plasma,
turbo, viridis, plasma,
],m
Max Bo
04/26/2023, 3:00 PMohhhhhhh!!!
Max Bo
04/26/2023, 3:02 PMthat's really interesting
f
Fil
04/26/2023, 3:02 PMyou could use d3.cross to make the matrix
m
Max Bo
04/26/2023, 3:06 PMsweeeet
🙌🏻 1
Max Bo
04/26/2023, 3:06 PMthanks folks!
f
Fil
04/26/2023, 3:09 PMawesome notebook
❤️ 1