« first day (1924 days earlier)      last day (2688 days later) » 

1:29 AM
I wish we could ping people from chat that haven't been here. Allow users to block that, of course, but it'd be handy in cases like now when I want to tell TunaMaxx to look at the improved edit I made on graphicdesign.stackexchange.com/posts/81300/revisions
 
I got the Gandalf 10 hour euro-trance disco nod thing :v @JohnB
 
 
1 hour later…
2:48 AM
@ZachSaucier you can ping people who edited the post
I don't think it autocompleted but it should still work
 
 
3 hours later…
5:25 AM
Right, but I don't want to leave comments hanging around if the person doesn't delete theirs afterwards
 
 
2 hours later…
7:34 AM
hey @ZachSaucier
What is agood easily configurable super simple blogging platform
 
 
1 hour later…
9:02 AM
I have a question, and I am not sure if it belongs here, or somewhere else on the SE network>
It is (i think) really technical, and I think if I asked on the wrong site it will not get a useful response.
 
Shoot.
If the folks in chat now can't help place it, hopefully someone will come along later who can.
 
(oh hi BESW, you are everywhere, no?)
(my old nick was oxinabox, we chatted a few times on RPG SE)
 
I thought I recognised that avatar. And yes, apparently I am everywhere.
What's new?
 
I'm thinking really hard about colors
 
We've got some colourful people here.
 
9:08 AM
So the Forbes color database characterises colors (of paints/pigments) using what is called a XRD (I think). Eg cameo.mfa.org/wiki/File:ChartImageLarge_2666.jpg
 
morning peeps
 
That's pretty cool.
 
Now, I don't really understand what XRD is (which perhaps would be a good question itself)
But my vague notion is that it captures the intensity of refelected electromagnetic radiation at different frequencies, for different input frequencies.
 
> X-ray powder diffraction (XRD) is a rapid analytical technique primarily used for phase identification of a crystalline material and can provide information on unit cell dimensions.
 
9:12 AM
So i think this is a more descriptive way of precisely describing a color, than as say a HSV triple (or RGB or CYMK etc).
 
Sounds like it's not for describing a color, but for describing a pigment.
 
The difference between a color and a pigment is that a pigment is the physical chemical that has a particular color?
 
Roughly, yes. The colour is the visual effect of the physical pigment.
afk a while gotta make dinner.
 
cya
So my question is (and I've been wondering this for a few months), is it possible to algorithmically convert from a XRD to a (A)RGB triple that I can display on a monitor, or ((A)CYMK) to print in a book.
And I am not sure if this a question for Graphic Design stack exchange, or Physics stack exchange, or maybe Photography stack exchange.
 
9:44 AM
@LyndonWhite yes and no
 
That is always the best kind of answer.
 
A spectral color composition is just that a spectrum of photon intensity
 
@LyndonWhite It's kinda asking to convert from a physical property to a digital one. I think Physics would bounce this to us and we to Physics... tough nut
 
RGB is just 3 spikes in the spectrum
Now it is possible to to integrate the spectrum and try to fit it in the RGB space
Hwoever quite a lot of colors will not succeed this transformation
Second, the color is not the same metamer. MEaning it will not react to surrounding light the same way
 
I was kinda thinking this, yes.
This is good. thank you.
A process that can fail (particularly if it fails in defined ways, at defined times) is still interesting to me (Probably more so that something that always works, eg RGB<->HSV)
 
9:55 AM
especially that orange is not possible to make
RGB to hosv is not a physical transformation no actual real color data is lost. your just moving form a abstract space to a nother abstract space.
You will need color management if you want to be even close
 
You know the difference between a glossy blue and a matte blue?
They're the same "color," and they'd look the same on a computer monitor. But "gloss" and "matte" are physical qualities with no digital equivalent.
They're easy to differentiate when printed, but identical in a light-only workspace.
 
@BESW they look the same under certain conditions
 
Also true.
 
/me nods
 
At the very least, you'd first have to feed the program an amazing number of examples and then see if it can extrapolate.
 
10:09 AM
Now that is entirely my kinda game
So the reason I am interested in this, is that I am interested in making a machine-learning-based natural language processing/generation system for generating colors based on descriptions (eg "Bright redish green") and generating descriptions from colors
I have about 4000 such named colors, and I was wondering about extending towards pigment databases as another source of informantion
 
Hrrm.
Pigment databases are descriptions of the physical qualities of a medium which, when viewed, is seen as a particular color.
RGB, CMYK, etc, are recipes for creating particular colors out of a certain physical medium (light, ink, etc).
The thing about pigments is that many of them are unrelated to the regimented color sets which modern recipe sets (like RGB and CMYK) rely on.
 
@LyndonWhite i dont think that works.
people as a agregate can not name colors
they do not agree on wht the exact hue is
 
There's a lot of different ways to make a cyan printer ink, or a blue light, and none of them have anything to do with Prussian Blue--whose intensity varies based on the size of its particles.
 
so going form uncertainnty to extreme exactness is not very useful
 
Indeed they do not agree on exactness, but that is fine.
NLP/Computational linguistics is full of issues like this.
 
10:20 AM
So, yeah, people can agree on what Prussian Blue is (a pigment produced by oxidation of ferrous ferrocyanide salts) but agreeing on what CMYK or RGB value replicates it? Hah.
There's literally not one answer. It's not subject disagreement, "Prussian blue" covers a wide span of blues.
 
sure -- you could say that Prussian Blue defines a probability distribution over a area of RGB space.
 
And depending on what it's suspended in, you can get oil paint, watercolour paint, acrylic paint... all of which is Prussian blue, has its own qualities of transparency and reflectivity, and varies in intensity depending on the size of the particles in the colloidal dispersion.
 
which refer to things that do not exist in monitor RGB space at all.
 
But which matter in physical space, which is what you're translating from. Don't forget The Dress.
It's a language problem wrapped in a physics problem wrapped in a programming problem, smothered by centuries of proprietary pigmentation techniques.
 
Loverly, isn't it?
It is such an incredibly interesting area.
 
10:28 AM
There are apps which let you point your phone camera at something and it'll tell you the colour of the thing.
One of my blind clients was delighted to discover these, but rather disappointed with the implementation.
The app tends to pick up ambient lighting cues more than object colors unless you put the camera so close to the object that it gets mostly shadow.
 
thanks for this discussion
 
To your original question: I think you're gonna have to break it down into bite-size chunks for multiple Stack questions, probably across multiple Stacks.
And please, keep me updated on the project, I'm very interested.
 
It is the challenges (even in RGB-Monitor space) that make it interesting.
As a subfield of linguistics, I think color naming is very dense in terms of having challenges (Eg ambiguities, pragmatic differences etc).
Which I suggest makes it a good area to pilot test systems for general NLP/ computational linguistics.
 
You may also find it difficult to translate into other languages.
Different cultures think of color differently, and their language reflects that.
 
Indeed. This is part of what I mean by "dense in problems".
 
10:41 AM
eg, English doesn't really go in for reflectivity as a primary color descriptor but some languages/cultures group colors by that, or by intensity, or...
(Hue is our primary descriptor.)
The most commonly known example is classical Greek, which considers the sky to be bronze-like and the sea to be wine-like.
 
 
2 hours later…
12:53 PM
@joojaa hey! If you're not looking to host it yourself, Medium is a good one
 
@ZachSaucier i can host
id prefer not to though
 
Medium it is
There are others but I think Medium is styled the best and has a better UX than the others I've tried that are free online
 
Thing is i need mathjax support
 
Ah
I use a blogging platform called Octopress which is built on Jekyll which isn't that hard, especially after the initial setup. It's a static website generator, meaning it only changes when I tell it to update on my local PC and then FTP the file changes over to the server
You can get MathJax support with Jekyll
Someone online says you get get MathJax for Tumblr too
 
1:11 PM
morning all
 
howdy
 
1:30 PM
evening
 
1:41 PM
hey there
@joojaa are you away from home?
or is it dark already in Finland? :P
 
full house today :)
 
royal flush, rather :P
 
@Vincent yes the sun set about 20 mins ago
oh seems to be 40 mins ago
so sun raised at 9:08 and set at 15:15 for a 6 hour day
 
for the web peeps among us, this is by far the most genious description of Service Workers I've ever read:
 
2:37 PM
Going home
 
 
3 hours later…
5:32 PM
@PieBie I've got a CSS/HTML question if you have a minute
nevermind, think I solved it browsing codepen. Had to put "clear: both;" onto a div
 
5:48 PM
Ah. Okido. Succes
 
2
Q: Adobe Illustrator Zoom around selected, not to screen centre?

ConfusedCommand (Cntrl) and = gives zoom in, in Adobe Illustrator. Is there a way to set it so that it zooms to the selected object rather than the centre of the screen?

 
6:10 PM
@Ryan maybe you should Google 'clearfix'
 
oh I think I may have one of those already from another page
Im such a good coder! hahaha not even a little
 

« first day (1924 days earlier)      last day (2688 days later) »