SP ABS Color Ring Scans

[Dianoda]

The only picture I can safely calibrate is the one Dianoda took.

Well, then shouldn't we just use his scans? If his are more accurate, then let's go with his scans.

we don't have the full color ring though

So, Dianoda hasn't photographed the rest of the PBT ring? Meh, I don't see the point in cropping out the chips if Dianoda's are going to represent the color more accurately. I guess we can wait until he's done.

I just have the ABS chips - no PBT.  For the ABS, I still need to shoot the gray chips - I was planning to take and upload these tonight (I've been out of town for the past week, just got back on Sun evening).  I'll probably reshoot the color chips at the same time as a comparison point to the last shoot - I'll be using the same lighting setup as before, just want to be sure that I am eliminating potential variables that I should be able to control (such as possible variations in metering and light output/color temperature of my strobes).

[mkawa]

i have to lend dianoda my pbt ring still. i might as well lend him the PC ring while i'm at it.

[Krogenar]

I'm trying to figure out why the PBT scans seem so overexposed. Maybe it's because I used more chips for the ABS scans, and it cut down on the light? Same scanner, same settings -- why such a difference? 

I think I'll rescan the ABS as well, just to get a second (third, with Dianoda's) set of values, and better scans for better crop-outs.

I wonder how the polycarb will look -- that's translucent, yes?

[mkawa]

it's trying to expose for middle gray, like any other exposure meter. so it takes a quick snapshot of reflectivity and computes an amount of light that will make the mean of the luminosities middle gray.

in short, scan the dark stuff separately from the light stuff.

[Krogenar]

it's trying to expose for middle gray, like any other exposure meter. so it takes a quick snapshot of reflectivity and computes an amount of light that will make the mean of the luminosities middle gray.

in short, scan the dark stuff separately from the light stuff.

(YEEEEEEEEEEEEEARRRRRRRRRRRRGHHHHHHHHHHHHHHH!!!!!!!!!!) (TIE fighter screeches over Krog's head)

 

I put the tiles on the scanner bed!    LOL

[Dianoda]

Ok, finished the reshoot of the ABS color chips and did the gray ABS chips as well.  Both images were shot using identical everything - same camera settings, same lighting setup from the same position, same post-processing steps, etc.  Here are links to the color corrected JPEGs (each JPEG file is about 12MB, btw):

http://dianoda.minus.com/muyOgFJnwJ8hW

The RAW files can be found on dropbox.

[Dianoda]

Oh, mkawa, just noticed your post asking about the color checker passport/i1 display pro - I bought the color checker stand alone, so no i1 for for me.  I use a spyder3 elite w/ argyllCMS+dispcalGUI for my screen calibrations, which I'm pretty happy with (the argyllCMS+dispcalGUI software is quite a bit more in depth than spyder's own software).  The color checker passport is great for studio work.

[jil_jil32]

It would be great if someone can send Dianoda those missing chips... ^~^

[mkawa]

Oh, mkawa, just noticed your post asking about the color checker passport/i1 display pro - I bought the color checker stand alone, so no i1 for for me.  I use a spyder3 elite w/ argyllCMS+dispcalGUI for my screen calibrations, which I'm pretty happy with (the argyllCMS+dispcalGUI software is quite a bit more in depth than spyder's own software).  The color checker passport is great for studio work.

hah. i bought the i1 without the passport. i suspect if we combine forces, we could be _unstoppable_...

[ijprest]

Just FYI:  Not sure how the scanning has progressed, but I've got a access to a ColorMunki spectrophotometer at work, if you wanted another set of calibrated samples.

[Krogenar]

Just FYI:  Not sure how the scanning has progressed, but I've got a access to a ColorMunki spectrophotometer at work, if you wanted another set of calibrated samples.

PM me with your address, and I'll send out the color chips (ABS) that Mashby sent to me a few months back. That work for everyone?

[mashby]

Just FYI:  Not sure how the scanning has progressed, but I've got a access to a ColorMunki spectrophotometer at work, if you wanted another set of calibrated samples.

PM me with your address, and I'll send out the color chips (ABS) that Mashby sent to me a few months back. That work for everyone?

Works for me.

[Krogenar]

Just FYI:  Not sure how the scanning has progressed, but I've got a access to a ColorMunki spectrophotometer at work, if you wanted another set of calibrated samples.

PM me with your address, and I'll send out the color chips (ABS) that Mashby sent to me a few months back. That work for everyone?

Works for me.

Package deployed.

[ijprest]

Got the package tonight, and borrowed the ColorMunki from work.  The scanning is actually going pretty quickly... did about 60 (roughly half) of the PBT chips over a couple hours, and I think I've got the process down pat.  Given the pace I'm going at, I should be able to get them all scanned over the weekend.

I'm sampling the chips from the smooth part... away from any texture or writing that might throw off the sample.  I'll do each chip five times, and average out the data.  The ColorMunki gives me raw data (samples of wavelengths from 380nm -> 730nm at 10nm intervals), as well as L*a*b* values.  I should be able to calculate sRGB values without any trouble.

[codyeatworld]

Just gonna say I've been using these scans for the recent group buys, and thanks a lot for doing this.

[ijprest]

OK, I've scanned all the PBT chips... went pretty quick.

• Raw data:  https://docs.google.com/spreadsheet/ccc?key=0As-LCp55OSYodE5TX3ZBYm5QZ3lqQktrb1JBSmZMcXc&usp=sharing
  
  • This is the raw data as pulled from the ColorMunki; LAB values, and samples at different wavelengths (380nm - 730nm, 10nm intervals).
  • I sampled each chip approximately five times, and I've included averaged results in with the raw data.
• Processed sRGB data: https://docs.google.com/spreadsheet/ccc?key=0As-LCp55OSYodEh4X0h1TnI3bS03Y295bDBmVnJpU3c&usp=sharing
  
  • In this spreadsheet I convert the LAB values into sRGB, using the formulas from Bruce Lindbloom's website.
  • Used a D65 whitepoint (because that's the default for sRGB), and correct sRGB gamma companding.
  • Feel free to double-check my work!  I checked a few samples against the online calculator, and was getting the same results.
  • If there is demand for other whitepoints (e.g., D50) or other RGB spaces (e.g., Adobe RGB), let me know and I can try to figure out the math.
  • Note that these colors will only be accurate if your monitor and/or graphics software is calibrated properly to display sRGB colors.

Of course, I've also added the colors to my online Keyboard Layout Editor.

I'll move on to the ABS chips tomorrow.

[Photoelectric]

Kroengar, if you do own any IMSTO white / beige PBT sets at some point, it'd be great if you could compare them to the ABS and PBT whites and beiges/grays to find the closest matches.  Mostly because IMSTO's sets only come in ANSI layouts, and I've been looking at KeycapsDirect for blank 1.5x, 1x caps and 7x spacebars.  There are many off-whites that look *similar* but even with color scans, it's uncertain which is the closest.  Would be best to have a color sample ring on hand to compare directly.

(For example, I ordered 4 1.5x modifiers in WFJ "white" PBT, and WFJ has been described as crisp clean white.  Upon arrival, I can see that it's in fact quite a bit creamier / more yellow than white Vortex caps and more cream than pure white PBT used for ZOMG dyesub caps (I thought those were WFJ, as well as that's what Portal dye sub caps used)--perhaps SP sent me wrong caps, I don't know.
P.S.: Well, as I was writing this, I've realized that indeed, my supposed WFJ modifiers are not correct, next to WFJ ZOMG dye sub caps, so I've sent an e-mail to SP customer service )

[Krogenar]

Kroengar, if you do own any IMSTO white / beige PBT sets at some point, it'd be great if you could compare them to the ABS and PBT whites and beiges/grays to find the closest matches.  Mostly because IMSTO's sets only come in ANSI layouts, and I've been looking at KeycapsDirect for blank 1.5x, 1x caps and 7x spacebars.  There are many off-whites that look *similar* but even with color scans, it's uncertain which is the closest.  Would be best to have a color sample ring on hand to compare directly.

No, I don't own any IMSTO anything, unfortunately. Many people have pointed out that the color rings from SP do differ over time. I encounter the same sort of thing in my business, where I use dyed rolls of fabric. The color is slightly different between "dye lots" and so I've got to keep all my sewing work within the same roll to keep it all consistent. SP mixes their colors, so we'll never be able to make perfect matches. But if people can come reasonably close that's good enough for us.

(For example, I ordered 4 1.5x modifiers in WFJ "white" PBT, and WFJ has been described as crisp clean white.  Upon arrival, I can see that it's in fact quite a bit creamier / more yellow than white Vortex caps and more cream than pure white PBT used for ZOMG dyesub caps (I thought those were WFJ, as well as that's what Portal dye sub caps used)--perhaps SP sent me wrong caps, I don't know.
P.S.: Well, as I was writing this, I've realized that indeed, my supposed WFJ modifiers are not correct, next to WFJ ZOMG dye sub caps, so I've sent an e-mail to SP customer service )

Yeah, so now you know from experience.  Think of this, though -- if the GH community can provide good, color-corrected scans of the various chips and materials, then it's easier (and more likely) that vendors will specify (or be asked to specify) what color chips they're using. But also, from what I have heard, SP will custom mix colors. So maybe the vendor said, "gimme WFJ with a dash of yellow, please."? The colors I find most interesting on the color rings so far have been YCC (soft yellow), YCA (translucent yellow) and RCU (eraser salmon). I feel like no one has done much with them. There's also a bewildering array of grays, LOL!

A tremendous thanks to ijprest, who is exactly the right kind of scientist (it's color science people, that's not a joke), computer scientist and keyboard aficionado. When these scans get into emulators and prototyping software, etc. -- like ijprest's -- then all kinds of awesome will happen. I'm hoping kaporkle, or someone with his skills, will arrive and make accurate rendering of new keycap colorways easier and faster.

[Krogenar]

Note that these colors will only be accurate if your monitor and/or graphics software is calibrated properly to display sRGB colors.
Of course, I've also added the colors to my online Keyboard Layout Editor.

That's awesome work! Checked out the PBT color scans on your editor, and yes, you should definitely calibrate your monitor if it's possible.

[ijprest]

Note that these colors will only be accurate if your monitor and/or graphics software is calibrated properly to display sRGB colors.
Of course, I've also added the colors to my online Keyboard Layout Editor.

That's awesome work! Checked out the PBT color scans on your editor, and yes, you should definitely calibrate your monitor if it's possible.

Yeah... color-management is hard. 

My monitor has an sRGB-mode, which makes a big difference.  I compared a few chips against the on-screen values, and they're reasonably close when I switch my monitor to sRGB... I chalk up the minor differences to my ****ty CFL lighting.  There's a reason the pros use monitor hoods!

But while I've got the ColorMunki here, I might try my hand at creating an ICC profile for my monitor and see if it gets even better. 

[korrelate]

Krogenar, you closet socialist: thank you for this link! I don't have to ask that many questions out here because every time I search for something somebody like you has already posted the answer. You rock!

[bpiphany]

[ Specified attachment is not available ]I had some scans uploaded at one point. I think they disappeared in the big file cleanse, and somehow I had deleted my copies as well =P So I scanned them again for fun... These are the ones I've got.

ABS


PBT

[Krogenar]

Krogenar, you closet socialist: thank you for this link! I don't have to ask that many questions out here because every time I search for something somebody like you has already posted the answer. You rock!

You're welcome! Be sure to thank all the other people who did the majority of the heavy lifting. mkawa, Dianoda, ijprest, and more. I started the thread, and then things got very quickly over my head in terms of technical sophistication.  Thank you bpiphany for the additional scans.

[ijprest]

I managed to scan all the ABS chips tonight... tomorrow night I'll double-check the data (e.g., make sure I typed the color-code correctly) and then post it online.

[ijprest]

OK, here's the ABS data (and links to the PBT data again, for good measure... though it hasn't changed).

• Raw data spreadsheet (ABS):  https://docs.google.com/spreadsheet/ccc?key=0As-LCp55OSYodDFiWDV4VkZhWWJ0R08yRFk5c25UZGc&usp=sharing
• Raw data spreadsheet (PBT):  https://docs.google.com/spreadsheet/ccc?key=0As-LCp55OSYodE5TX3ZBYm5QZ3lqQktrb1JBSmZMcXc&usp=sharing
  
  • This is the raw data as pulled from the ColorMunki; LAB values, and samples at different wavelengths (380nm - 730nm, 10nm intervals).
  • I sampled each chip approximately five times, and I've included averaged results in with the raw data.
• Processed sRGB data (ABS): https://docs.google.com/spreadsheet/ccc?key=0As-LCp55OSYodFdTdnpoWG1iYXM4aE1EaUJQRnhSOEE&usp=sharing
• Processed sRGB data (PBT): https://docs.google.com/spreadsheet/ccc?key=0As-LCp55OSYodEh4X0h1TnI3bS03Y295bDBmVnJpU3c&usp=sharing
  
  • In this spreadsheet I convert the LAB values into sRGB, using the formulas from Bruce Lindbloom's website.
  • Used a D65 whitepoint (because that's the default for sRGB), and correct sRGB gamma companding.
  • Feel free to double-check my work!  I checked a few samples against the online calculator, and was getting the same results.
  • If there is demand for other whitepoints (e.g., D50) or other RGB spaces (e.g., Adobe RGB), let me know and I can try to figure out the math.
  • Note that these colors will only be accurate if your monitor and/or graphics software is calibrated properly to display sRGB colors.
• Raw data .CXF (ABS):  https://drive.google.com/file/d/0B8-LCp55OSYoZ29hNnpsQkNFc0E/edit?usp=sharing
• Raw data .CXF (PBT):  https://drive.google.com/file/d/0B8-LCp55OSYoT0ZPVlJTTjl1S2c/edit?usp=sharing
  
  • This is the "palette" file generated by the ColorMunki software.
  • Included for completeness.  It's probably not useful for most folks, but maybe your graphics software can read it.
  • Note: This version has all the original samples, but doesn't include the averaged colors... this probably limits its usefulness further.

I've also added the colors to my online Keyboard Layout Editor.

Edit: I found & fixed a couple places where I had transposed a couple characters in the color-names, and one instance where I had mis-read an 'L' as a 'C'.  Plus, I discovered that we have two VBQ chips.

[ijprest]

P.S., I'm not sure how accurate the YCA (glow-in-the-dark) samples are.  It was glowing for a few seconds after each sample I took. 

[korrelate]

Krogenar, you closet socialist: thank you for this link! I don't have to ask that many questions out here because every time I search for something somebody like you has already posted the answer. You rock!

You're welcome! Be sure to thank all the other people who did the majority of the heavy lifting. mkawa, Dianoda, ijprest, and more. I started the thread, and then things got very quickly over my head in terms of technical sophistication.  Thank you bpiphany for the additional scans.

Roger that!

mkawa, you know I already love you (and your heart stickers) so of course, of course: many thanks!!

Dianoda, ijprest, et al.: big thanks to all of you as well. I haven't had a chance to interact with any of you yet, but I'm sure that will happen sooner or later.

And to all:

Thanks for being interested in the things that you're interested in and thanks for sharing the info! You guys all make me very glad I decided to join this forum. Feel that? That's good karma being sent your way!!

[Krogenar]

Ok, I'm working on getting some of these scans cropped out so they can be used. I've figured out a way to crop them with rounded corners, so they look a bit more like the tiles. Now, if only I can figure out how to tag each one with some of this new data. Here's what some of the new cropped jpeg will look:

[ijprest]

What sort of tagging were you considering?  Something to allow for programmatic access?

Putting stuff into the image metadata is an option.  The Exif metadata format supports lots of different bits of metadata, and PropertyTagExifMakerNote or PropertyTagExifUserComment fields can be used for just about anything, if none of the other fields make sense.

[Krogenar]

What sort of tagging were you considering?  Something to allow for programmatic access?
Putting stuff into the image metadata is an option.  The Exif metadata format supports lots of different bits of metadata, and PropertyTagExifMakerNote or PropertyTagExifUserComment fields can be used for just about anything, if none of the other fields make sense.

Hrm... I hadn't actually thought about using IPTC or EXIF to put the data right into the file itself.
I was thinking something more like this:



Putting that information in the metadata of the file would be awesome. I'm trying to think in terms of what would be most useful to someone designing a color scheme, or putting together a groupbuy. Ultimately, if someone could use this information to create better CG renderings, that would also be fantastic. I had also pictured a PS or Illustrator swatch that could be loaded while designing, so people could have a palette of ABS chip colors to choose from. Again, we can't mention the profiled monitor caveat often enough!

Your integration of the swatches into your colorscheme designer is looking awesome. How difficult would it be to have the swatches (the cropped photos) appear in the design window? I'll kludge something together to show you what I mean.

[Krogenar]

I'm finally giving this thread some needed attention. OP updated.

[Krogenar]

For the purposes of this discussion, we'll be using the "Straight Up House Tully" color scheme. *ahem* 

Here's the editor right now:



Could we do something like this? When I hover over your palette, the tooltips indicate which color chips have been used. But after I select them, I really need to take notes -- if I don't recall what I used, I have to select a few until I find what I had used. It would be so useful if whatever colors I did select from a color chip gamut, showed up somehow. Please don't take this is as a harsh criticism at all -- I feel a bit like you managed to levitate the Statue of Liberty and now I'm asking you to do it a little bit higher, as I still cannot see entirely up her toga. Here's my kludged screenshot:



What do you think? Possible? I would love to see more designers using the tool you made, and then being able to output some view that shows all the basic information, screenshot it, and then post it. Most of the GB's out there end up with a screenshot(s) on the OP that give people all the basic info on the color scheme -- if your tool could make that easier to output, that would be great.

Also, Santa, what about an online color chip "ring", that users could use to rifle through the chips, select some to "pull out" and then it would use the higher-res images? Also, I want a Red Ryder bb gun and I totally won't shoot my eye out.

[ijprest]

I feel a bit like you managed to levitate the Statue of Liberty and now I'm asking you to do it a little bit higher, as I still cannot see entirely up her toga.

Heh... that's some vivid imagery. 

Something like this is certainly possible... it's a question of best presenting the data in the UI, and having a workflow that makes sense and is easy to use.    (And fits in with the other features I have planned.)  One consideration is that you could (theoretically) use different colors on each keycap... so presenting the information in a useful way becomes tricky.

A few ideas:
1. To solve the "I don't remember what color I picked" issue, I could just highlight the current color in the palette (sort of like matt3o's color-scheme picker draws a thick border around the current selection).

2. I plan to implement (at some point) the concept of named keycap prototypes that specify various attributes.  You could have a "alphas" prototype and a "mods" prototype (or whatever; system would be flexible), apply the prototypes to the correct keys, then *quickly* test color schemes, etc., just by changing the colors on the prototype caps.  I'm not sure exactly what the UI will look like, yet, but I've been thinking about something like this mockup:


... so another possibility would be to show the scanned chips beside each prototype in the list.

3. Another thing I plan to implement is the ability to add text- or image-based notes/decals to the layout...
-- Lots of uses for this... logos, depicting LED clusters, labelling "row number" information, etc.
-- This idea could be easily extended to allow "color swatch" decals.
-- This would be best (IMO) for taking screenshots to use in GB posts.

[ijprest]

OK, here's the ABS data (and links to the PBT data again, for good measure... though it hasn't changed).

P.S., There was no NN (black) chip on the chain, but I ordered a blank NN-colored key from SP and scanned it in... the spreadsheets and such that I linked above have been updated.

[Krogenar]

A few ideas:
1. To solve the "I don't remember what color I picked" issue, I could just highlight the current color in the palette (sort of like matt3o's color-scheme picker draws a thick border around the current selection).

That would work -- this is really a great tool you've made.

2. I plan to implement (at some point) the concept of named keycap prototypes that specify various attributes.  You could have a "alphas" prototype and a "mods" prototype (or whatever; system would be flexible), apply the prototypes to the correct keys, then *quickly* test color schemes, etc., just by changing the colors on the prototype caps.  I'm not sure exactly what the UI will look like, yet, but I've been thinking about something like this mockup:
(Attachment Link)

... so another possibility would be to show the scanned chips beside each prototype in the list.

It sounds like CSS for keycap color schemes! (Inheritance, etc.) -- I think that's a great idea. Matteo's tool does this already -- alphas vs. mods, but over time I've realized that schemes also play with other color "models" -- alternating function row cluster colors, does the tab key get the mod color, or the alpha color? Being able to select/create a grouping of keys would be pretty awesome.

3. Another thing I plan to implement is the ability to add text- or image-based notes/decals to the layout...
-- Lots of uses for this... logos, depicting LED clusters, labelling "row number" information, etc.
-- This idea could be easily extended to allow "color swatch" decals.
-- This would be best (IMO) for taking screenshots to use in GB posts.

Do you think you could put together a way of creating an online (or software-based) color chip keyring?

[Matt3o]

I don't remember if we already talked about this. But colors are not constant, they slightly change between batches. So what they today call VAT might be different from VAT a year from now.

[Krogenar]

I don't remember if we already talked about this. But colors are not constant, they slightly change between batches. So what they today call VAT might be different from VAT a year from now.

Yes, that's been pointed out many times. Also, monitors have different temperatures, too -- so even if everything were perfectly calibrated, it wouldn't be 100% accurate. You could make the argument that SP shouldn't even sell color chips for the same reason -- they'll be different eventually from what you've get. And yet, people still buy the color rings.

[Matt3o]

I mean that... just don't go paranoid about calibration a perfect rgb convertion because it won't last

[Dianoda]

So... I imaged the SP PBT and polycarb rings this evening, RAW files can be found on the GH dropbox, here are the jpegs (sRGB colorspace):

http://min.us/muyOgFJnwJ8hW

[jacobolus]

The ColorMunki gives me raw data (samples of wavelengths from 380nm -> 730nm at 10nm intervals), as well as L*a*b* values.  I should be able to calculate sRGB values without any trouble.

Oh, holy ****, this is awesome. Somehow I didn’t realize that these color chips had been measured with a spectrophotometer. Thanks so much. Yay data to play with.

Why is this in the “post wiki workshop” section?

[Krogenar]

So... I imaged the SP PBT and polycarb rings this evening, RAW files can be found on the GH dropbox, here are the jpegs (sRGB colorspace):

http://min.us/muyOgFJnwJ8hW

Thanks Dianoda! I need to get to work on cropping these things out. My hope was to crop out rounded sections, to match the actual chips, but it's ... well, very tedious. Anyway, thanks for the scans Dianoda!

The ColorMunki gives me raw data (samples of wavelengths from 380nm -> 730nm at 10nm intervals), as well as L*a*b* values.  I should be able to calculate sRGB values without any trouble.

Oh, holy ****, this is awesome. Somehow I didn’t realize that these color chips had been measured with a spectrophotometer. Thanks so much. Yay data to play with.

Why is this in the “post wiki workshop” section?

Where should it be, if not here? The working plan has been that since the wiki code requires too great a learning curve, that all the information we compile would be in this forum, until a more long-term solution is found. Glad the scans helped you!

[Azteca]

Great work 

Any chance someone knows the PBT color used in darker Realforce keyboards, and better if someone knows the ABS color equivalent (or closest).



Cheers