Pages: [1] 2   Go Down

Author Topic: Certification for RGB devices?  (Read 9781 times)

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Certification for RGB devices?
« on: April 17, 2012, 01:40:23 pm »

Hello, what is suggested workflow for RGB device certification?
I'm unaware of any international standards but perhaps there is some tutorial, workflow etc. for homemade way of doing this?
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Certification for RGB devices?
« Reply #1 on: April 17, 2012, 02:29:45 pm »

Can you be more specific as to what you want?

One can examine the measured output of an RGB device compared to what the profile predicts.
One can compare how the same RGB values output multiple times to define device stability and consistency.
One can compare how the same RGB values differ over the sheet being printed.

All of the above and probably more.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #2 on: April 17, 2012, 03:10:40 pm »

I was looking for a way to evaluate RGB profiles like a test chart with some "standardized" values, like mediawedge 3.

My current workflow is:

I have opened Media wedge 3 CMYK with Measuretool, and saved LAB data to text file.
I think it is possible to use any test chart or am I wrong?

Then I used colorthink worksheet to load this LAB data and apply my device RGB profile. This gave me RGB/LAB data with my device profile applied to it. I save it as colorlist. As I understand this converts colors and makes sure non printable colors are not there right?
I checked for out of gamut colors by viewing/softproofing the target I made from colorthink colorlist data in PS with my device profile, no out of gamut colors were there.

Then I have made a test chart from these RGB values with Measuretool, open print from PS convert to profile (my RGB profile) print with no color management as I allready converted my test chart to my device profile.

Then I opened my measured data to compare in measuretool and in colorthing worksheet. To my amazement some patches were 4 some 2.5 dE.
« Last Edit: April 17, 2012, 03:13:54 pm by smilem »
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Certification for RGB devices?
« Reply #3 on: April 17, 2012, 03:21:46 pm »

I was looking for a way to evaluate RGB profiles like a test chart with some "standardized" values, like mediawedge 3.

You can use the target you used to build the profile and compare those Lab values to the Lab values the profile actually produced.
You can build your own target too, useful if you want to check more neutrals or if you want to check ‘house’ colors (specific logo colors, skin tones etc).

I’m not sure why you are messing with the Media Wedge CMYK target.

Once you decide what colors you want to compare, you’re going about it the right way using ColorThink. However, ColorThink does NOT convert using Black Point Compensation! So if you want to factor that in (because you are converting using Photoshop), then you’ll have to do all conversions in Photoshop to the patches, sample down (Nearest Neighbor), then load that into ColorThink and build a ColorList there. Then you can compare the reference values to the output values to get a dE report. When you convert in Photoshop, you’ll want Dither off. You also have to decide what rendering intent you want to test. If you always use a Perceptual rendering, you’ll use that to convert the patches using the profile you want to test. Using Absolute gives you a better set of values but if you never use that to convert the data, what’s the point?

Again, you need to consider what you want to run through the profile to compare. You may want to use the iStar target. See: http://www.aardenburg-imaging.com/
Mark hangs out here, he might see this and comment on it’s further usage. If you are using color patches to evaluate that are known to be outside the gamut of the profile, well that’s kind of a waste. So the patches you select can play a role in the dE report you get from ColorThink.

This may fill in the holes I missed:

http://digitaldog.net/files/How_to_compare_deltaE_of_ColorLists.pdf
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #4 on: April 17, 2012, 04:16:59 pm »

Thanks for detailed information on workflow, I have a question:

When I open my target in photoshop (convert to output RGB profile) the Lab values of color picker are different than colorthink shows when I make colorlist from same RGB values. My rendering intent is absolute so BPC is not interfering here.

Colorthink LAB 38.94 19.04 16.82
Photoshop LAB 35 28 30
RGB values are 122 94 69

What LAB to use as reference?
« Last Edit: April 17, 2012, 04:27:29 pm by smilem »
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Certification for RGB devices?
« Reply #5 on: April 17, 2012, 04:21:43 pm »

Do you have the PS eyedropper sampling set to 1x1?

Also again, ColorThink isn’t using ACE although I doubt whatever CMM is used would differ that much. And dither should be off.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #6 on: April 17, 2012, 04:29:47 pm »

Do you have the PS eyedropper sampling set to 1x1?

Also again, ColorThink isn’t using ACE although I doubt whatever CMM is used would differ that much. And dither should be off.

The photoshop color settings was set to relative, and the color picker used that to convert the data. I changed it to absolute and lab matches with colorthink
« Last Edit: April 17, 2012, 04:37:36 pm by smilem »
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #7 on: April 18, 2012, 07:08:24 am »

I’m not sure why you are messing with the Media Wedge CMYK target.

That is because PM5 can generate cool reports by itself without third party software, I tried to generate report for RGB device. But somehow I only get error on generating it.
Logged

Scott Martin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1315
    • Onsight
Re: Certification for RGB devices?
« Reply #8 on: April 18, 2012, 10:11:20 am »

That is because PM5 can generate cool reports by itself without third party software, I tried to generate report for RGB device. But somehow I only get error on generating it.

I've requested RGB Quality analysis but it's low on the priority scale. The ColorChecker proof and real world printing results are perhaps the next best thing for now.
Logged
Scott Martin
www.on-sight.com

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #9 on: April 19, 2012, 05:41:55 pm »

What about IT8.7-2 target, references are made by reading printed target on a true rgb device (digital lab), so because digital labs doesnt have widest colorspace but are accurate it is fair that comparing any other RGB printer to these colors is also fair.

I tried making rgb image (srgb.icc as source) from it8.7-2 lab reference and to my amazement best glossy papers like canon pro, and pro printers like ipf8300, canon pro-1, 9000 mk II, can reproduce all colors except last very very black patch acording to photoshop absolute intent. no gamut warning.

Less professional papers, matt papers on same printers (manufacturer profiles were used) does not reproduce all colors. Making any synthetic chart ant converting it to printer profile in ps allows to track icc accuracy, but no standardization there not even possible to compare what a device should reproduce.

Hasn't anyone noticed a need for such standardized target ? Why there is no standardized target to test RGB printers?
« Last Edit: April 19, 2012, 05:44:54 pm by smilem »
Logged

Scott Martin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1315
    • Onsight
Re: Certification for RGB devices?
« Reply #10 on: April 20, 2012, 09:16:06 am »

What about IT8.7-2 target...

A dying breed! I think it's all Colorcheckers from here on out.

Hasn't anyone noticed a need for such standardized target ? Why there is no standardized target to test RGB printers?

Not really. Offset press work has a hugely obvious need for specifications like SWOP, GRACOL and FOGRA. With silver halide printers, inkjet printers, etc you just calibrate and profile them and you're off and running. You get what you get. Super easy in comparison. There's not a whole lot of guess work (ie should we calibrate it to this or that specification). On these devices the calibrated state is optimal for that process and there's no variably to it. Coming up with specifications like SWOP is super tedious work that would need to be done by an independent 3 rd party. 

Now we could come up with a gamut rating scale for both silver halide  and inkjet printers and potentially even have an estimated average value to compare to. That might be fun to work with. That and a graph that shows device drift over time. This is the kinda of thing that might show up in i1P years from now after more pressing issues have been resolved. Or perhaps we'll see it sooner in other software.
Logged
Scott Martin
www.on-sight.com

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #11 on: April 22, 2012, 09:12:32 am »

A dying breed! I think it's all Colorcheckers from here on out.

I don't think so, look at attached (PS softproofing screenshots) files the comparison shows brother MFC (small gamut) cheap matt paper, and Canon IPF8300 (large gamut) paper is not cheap ant it is glossy too.

As you can see colorchecker does not allow to see and judge about printer color reproduction, all colors are printable on low gamut printers, while this is good for checking color accuracy it can be done with any colors converted to printer ICC profile.

I was taking about target that would allow to standardize RGB printer print quality, and it is not hard to import data for excel for dE calculation (formulas are available) so custom patches can be evaluated not harder than Fogra charts. It is possible to set custom tolerances for paper color patches and primary solids, grayscale etc.

It also seems possible by using IT8 chart I can chek what printing settings are better, I soft proofed OEM profiles and it clearly shows that a higher print setting on same paper reproduces all patches, when lower print quality setting can't reproduce deep blacks, and some patches.
« Last Edit: April 22, 2012, 09:23:04 am by smilem »
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #12 on: April 22, 2012, 09:28:15 am »

Here is the print quality comparison I talked about using IT8.7-2 target the colorchecker can't to do this.
Logged

Scott Martin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1315
    • Onsight
Re: Certification for RGB devices?
« Reply #13 on: April 22, 2012, 11:03:58 am »

I said colorcheckers not colorchecker. :-] I expect to see this family moving forward and question the future availability of silver halide IT8 targets. Surely someone can do better than silver halide IT8 targets.... Yes the 24 patch colorchecker has some serious limitations. Repeat your test with the Colorchecker SG though, and see what you come up with.
Logged
Scott Martin
www.on-sight.com

chrismurphy

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 77
Re: Certification for RGB devices?
« Reply #14 on: April 22, 2012, 12:56:32 pm »

The IT8.7/2 is an RGB target without any corresponding XYZ or LAB values. Any reference file you associate with it is an arbitrary choice, and biased toward film. I don't see how that actually gains you anything. While the Color Checker is a subjective, not arbitrary, choice, one does so in part because it has objective ways of evaluating the results in a comparison. You can do it by comparing measurements. Or you can whip out an actual color checker, along with a printed version, and use your eyeballs as the comparator

I think the failure in understanding what you want to do stems from not understanding the difference between commercial printing metrics, and photographic printing metrics. There are such things in commercial printing because of the need for film or file portability: film before CTP platemaking and today for modern workflows it's file portability. These are files transported around that are targeted for a specific kind of printing. In photography, we don't run around with "strike plates" anymore, trying to get them printing in a variety of locations. We tend to work using non-destructive workflows, abstracted from the particular requirements of the print making process.

The point is to print "the image" as best as it can be printed regardless of media or printer used. That's not at all the same requirement for printing presses. There, it's not about printing the best each press can print. It's about printing consistently.
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #15 on: April 23, 2012, 07:58:13 am »

The IT8.7/2 is an RGB target without any corresponding XYZ or LAB values. Any reference file you associate with it is an arbitrary choice, and biased toward film. I don't see how that actually gains you anything.

Wrong, the values are defined by ISO 12641-1997 standard, you can read them here:
http://www.docstoc.com/docs/25691578/?ct=40&utm_source=docstoc&utm_medium=email&utm_term=Registration+-+Doc&utm_content=Registration+Confirmation+With+Doc+-+DP+-+v2&utm_campaign=Registration&alt=985f1cd6-e30b-4937-91ae-37d268d6db8e

Or you can buy whole standard if you want. Values are not in LAB format but you convert it to LAB without problems.


While the Color Checker is a subjective, not arbitrary, choice, one does so in part because it has objective ways of evaluating the results in a comparison. You can do it by comparing measurements. Or you can whip out an actual color checker, along with a printed version, and use your eyeballs as the comparator

You can do this with IT8.7-2 too printing the target is far cheaper than buying the colorchecker SG, in fact there are numerous sources for IT8 targets that for reasonable prices.

Colorchecker SG does not have 22 grayscale paches to begin with, it also lacks the CMYK RGB step values available in IT8.

The point is to print "the image" as best as it can be printed regardless of media or printer used. That's not at all the same requirement for printing presses. There, it's not about printing the best each press can print. It's about printing consistently.

Agreed.
Logged

chrismurphy

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 77
Re: Certification for RGB devices?
« Reply #16 on: April 23, 2012, 12:32:33 pm »

Wrong, the values are defined by ISO 12641-1997 standard,

The paper you cite does not say the values are defined in a CIE space. Even if they were, it's absurd, because as anyone who has actually ever looked at targets conforming to IT8.7/2, they are not even remotely the same amongst each other. The paper you cite rather clearly demonstrates this fact.

Quote
Or you can buy whole standard if you want. Values are not in LAB format but you convert it to LAB without problems.

OK well if you consider the choice of reference not a problem. I'd consider it a rather significant problem if your goal is to avoid arbitrary decisions and results. You do not have a single reference available here. Period. You have nothing like a TR006 (which correlates CMYK to LAB) that's agreed upon by your industry, nor the equally important issue of tolerances. You can't have the same tolerance for all colors, you realize why don't you?

Quote
You can do this with IT8.7-2 too printing the target is far cheaper than buying the colorchecker SG, in fact there are numerous sources for IT8 targets that for reasonable prices.

YOu're not buying an IT8.7/2 either. You're printing it. So why print that target as your control slug to verify the correct output of an "RGB" output device, rather than either the values in the original Color Checker, or Color Checker SG, or some other target?


Quote
Colorchecker SG does not have 22 grayscale paches to begin with, it also lacks the CMYK RGB step values available in IT8.

Have you ever seen press process control use 22 grayscale patches? Have you ever seen it use more than 4? Why do you want more? I can imagine some good reasons why, but unless you're really clear about why, you're not really going anywhere.

The fact of the matter is, your industry doesn't have standardized print behaviors like the print industry does. So anything you come up with is somewhere between arbitrary and subjective.
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #17 on: April 24, 2012, 12:23:11 pm »

I have found this site http://www.colorcheck-online.de what do you think about their proposed test chart, methods of quality check for RGB printers?

Logged

chrismurphy

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 77
Re: Certification for RGB devices?
« Reply #18 on: April 24, 2012, 01:35:29 pm »

There isn't enough information on the site to answer the question. I couldn't readily find a way to download the test chart. The screen shot implies there's sRGB, red, green, blue and if those are sRGB primaries they're going to be out of gamut for a printer on most any media, so that's going to cause huge errors that are almost meaningless. When I click on Tolerances I'm asked to login. So I have no idea what the tolerances are. It might be a reasonable and helpful product for those looking to automate and track process control for fine art printing. And that's probably reasonable.

But the human visual system is a great color comparator. Most people can print a reference image on non-OBA media, stick it appropriate storage, and reprint periodically and compare. *shrug*
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 297
    • Color Management Services
Re: Certification for RGB devices?
« Reply #19 on: April 24, 2012, 01:56:34 pm »

Not all website is available in English, but if you click the English flag then on the right you click "Download" you get test charts.

They are available in various color spaces, the target has smart design:

Row A evaluates gray balance, last 4 patches evaluate Shadow details
Row B evaluates printer accuracy
Row C defines the gamut

They offer 5 uploads for free tests for any so I think I will just upload and see what I get reported.

There is demo account: username: demo password: demo
« Last Edit: April 24, 2012, 02:20:35 pm by smilem »
Logged
Pages: [1] 2   Go Up