@EineGeschichteDerStadtKoelnPod @Lokal_K ups 🙃 … das Lokal K ist ein lokaler Community-Raum in Ehrenfeld, jeden Donnerstag treffen sich dort Aktive der „hiesigen“ #Wikipedia-Community, aber der Raum wird auch von verwandten Geistern wie #OSM, #oklab u.a genutzt. Vor Ort wird an Projekten gearbeitet, gequatscht und zusammen gegessen. https://de.wikipedia.org/wiki/Wikipedia:Lokal_K
#ReleaseThursday This is actually the 2nd bunch of #ThingUmbrella releases since the weekend (which I completely forgot to notify about). The main highlights are:
https://thi.ng/color - added #Oklch support, update #Oklab conversion matrices (to latest spec), updated/improved/fixed CSS color parsing & serializations (more input formats, wider support for using percentages in more places), added serialization options for CSS Color Module Level 4, more docs
https://thi.ng/hiccup-markdown - major new version w/ completely new & extensible parser to transform #Markdown (with custom extensions[1]) into a proper abstract syntax tree and then transform elements via user defined handlers into arbitrary data structures. The default handlers generate a https://thi.ng/hiccup tree. More info in the package readme and the readme of this interactive parser & live preview demo: https://demo.thi.ng/umbrella/markdown/
[1] The syntax extensions include blocks of arbitrary metadata attached to _any_ blocklevel element (e.g. headings, paragraphs, tables, lists etc.) and custom typed blocks of content
As usual, please see changelogs for other updates:
https://github.com/thi-ng/umbrella/blob/develop/README.md#latest-updates
Happy Coding! 😍
#releasethursday #thingumbrella #oklch #oklab #markdown
Advice from watching @svgeesus's "Escaping the sRGB Prison" talk: if you want to polar color mixing use #Oklch, if you want linear color mixing use #Oklab. See also: https://developer.mozilla.org/en-US/docs/Web/CSS/color_value/oklab
Alright, now we're cooking. #matplotlib is not that bad :)
#matplotlib #animation #math #color #oklab
I also like how some patterns start to emerge. Here for instance the "a" component (green/red) is almost perfectly aligned linearly to the "b" component (blue/yellow).
Note: if you're confused about the 3D perspective view, that's because of a bug in #matplotlib (https://github.com/matplotlib/matplotlib/issues/24767)
#matplotlib #color #research #oklab
I'm now able to make all kind of combinations of the following params:
- honor or not the weights of the colors
- use variance or simpler heuristic for box selection
- normalize or not the variance
- refine with k-means with arbitrary iterations
- work in sRGB, linear or #OkLab
I'm tempted to replace the median-cut with a PCA as first pass, add an axis-variance parameter, and maybe adjust box selection accordingly).
Then I'll need to generate and study all kind combinations…
I'm currently experimenting with palette quantization algorithms. Here is a fig with 2 palettes (represented in 3 color spaces each): the reference computed in sRGB (top), and the new in perceptual (bottom, #oklab) .
For some reason in the new palette (bottom row) the colors are way more sparse with zebra gaps, as if there were less colors (this is not the case). As a result the original picture quantized with the new palette suffer from terrible banding.
I'm annoyed, but not defeated.
I just discovered what I think is THE BEST color picker on the internet**. Seriously.
❤️❤️❤️
🎨🎨🎨
🌈🌈🌈
📊📊📊
🍾🍾🍾🍾🍾❤️❤️❤️❤️❤️
https://oklch.evilmartians.io/#45.3,0.108,102.285,100
AND, as if that wasn’t enough — they wrote an incredibly helpful and in depth blogpost as well.
https://evilmartians.com/chronicles/oklch-in-css-why-quit-rgb-hsl#oklch-vs-oklab--lch-vs-lab
These use #OKLab and #OKLCH, which are for many situations the best colorspace to use (especially for color on the web).
**technically, “my discovery” was a Git update. But, it was a most exciting Git update 🥳
Yay, I got my fixed point #OkLab implementation working and the precision is much better than my expectations.
Comparing it against the official float/libm implementation on the 16M+ sRGB colors raises a maximum error of 0.000047.
It's unfortunately slower right now: 526.0ms ± 4.9 ms vs 816.2ms ± 4.1 ms
Quick color/programming question:
Does anyone know of a good tool/website for quickly comparing a block of hex codes (or similar) against each other?
Ideally I could paste in 5, or 8, or 50 #Color codes…and get as output a comparison matrix of differences as #DeltaE2000 or #OKLab
I’ve previously used a behemoth #Excel + Google sheets to do this, but it was absurdly manual to set up all of the lookups…and there have to be others who have already made something like this?? #Colorjs
#Colorjs #excel #oklab #deltae2000 #color
For people who work with #OKLab, #Color, #SVG, or #ColorProfiles...Are you familiar with Mark Fairchild and Yuta Asano's work on characterizing Categorical Observers?
https://onlinelibrary.wiley.com/doi/10.1002/col.22493
I'm trying to see if anyone has experience or advice on how to implement it.
#colorprofiles #svg #color #oklab
@Justin_Lind @havegum I think @gka had a JS library for this, but the name escapes me at the moment. Dunno if it's been updated with newer color spaces like #OKLab and #OKLCH
@havegum
#OKLab is currently the best (I believe, with #LAB being the older and maybe second-best place to operate in) #ColorSpace in which to make all of these transitions & color ramps.
Here are two useful sites, which can hopefully get you headed in the right direction:
A mmber of our #oklab analyses the state of #opendata portals in Germany, and developed a processing for the metadata of #ckan portals, aggregated and visualises this in #tableau.
You can have a look here:
https://public.tableau.com/app/profile/michael.riedm.ller/viz/bersichtOpenDataPortale/Dashboard2
And here is a peek into a part of the preprocessing chain necessary, implemented with #knime:
#oklab #opendata #ckan #tableau #knime
#Kiel bekommt ein Open Knowledge Lab! Mach mit! #OKLab
https://www.webmontag-kiel.de/2020/02/10/kiel-bekommt-ein-open-knowledge-lab-mach-mit/
#Kiel bekommt ein OK Lab!
Auftaktveranstaltung für das #OKLab Kiel ist am 17. Februar 18 Uhr in
den Räumen von opencampus.sh in der Kuhnkestraße 6 im Wissenschaftspark.
Es gibt ein wenig Input zum Thema "OK Lab" und #OpenData. Anschließend wollen wir überlegen, was wir im OK Lab Kiel machen wollen. Also gerne schon Ideen für Projekte mitbringen.
Hier kann man erfahren, was ein OK Lab ist: https://www.codefor.de/ueber/