Luminous Landscape Forum
Raw & Post Processing, Printing => Colour Management => Topic started by: Alessandro_V on December 20, 2013, 11:06:45 am
-
Hello,
I am writing to you to know if you had any past experience with home made calibration chart.
I explain myself better...
I thought it would have been a good idea to make my own calibration charts, to be used to create my camera ICC profiles, so to have some special colors and more patches than the charts you could commonly buy (also due to the costs...).
My idea is to use Argyll together with a spectophotometer to process everything, from the creation of the reference files to the final ICC.
It should be too difficult but still I wanted to point out my process to have your feedbacks and critics:
- I would generate a chart with Photoshop, creating my own template, and chosing the colors to be used in an arbitrary manner (obviously including all the necessary colors to make a perfect ICC profile in addition to the special ones).
Also, I would leave some empty spaces so to manually insert later on some special color patches that can't be normally printed with a common printer. - I would print my chart using my printer (I have a very good printer) and I would insert the missing ones by hand (basically cropping the new patch and gluing them to the original chart plus the addition of a light trap for the reference black)
- I would create a *.cht file in this way (http://www.argyllcms.com/doc/cht_format.html)
- I would read the chart with the spectophotometer to create the *.ti3 (chartread) file and convert it to *cie (spec2cie) file for reference.
- Finally I would shoot the chart with the camera and use the file to create the *.ICC profile (colprof)
Thanks a lot in advance for your help!
My best,
Ale
-
It may be a good idea if you want to take pictures of prints from your printer.
In all other cases it's a waste of time and ink.
-
Why? :-[
-
Real world objects have spectral properties that are different than target patches printed on a printer.
-
there is a big proponent of home made charts - Ben Goren ( http://www.trumpetpower.com/photos/Exposure#Making_a_chart ) , google him @ lists.apple.com/archives/colorsync-users
-
Thanks a lot Vladimirovich for that link!
I am a bit confused... Isn't the IT8, that usually has a better reputation for building ICC profile rather than the ColorChecker24, simply printed (obviously on various supports...)?
If I print a chart and I complete it with some other pigment, hans applied, I don't see why I should get bad results if I do everything properly...
Thanks a lot!
-
Thanks a lot Vladimirovich for that link!
I am a bit confused... Isn't the IT8, that usually has a better reputation for building ICC profile rather than the ColorChecker24, simply printed (obviously on various supports...)?
If I print a chart and I complete it with some other pigment, hans applied, I don't see why I should get bad results if I do everything properly...
Thanks a lot!
Undeserved reputation. Nice for CMY photo dyes like scanning or camera copying color photos.
At best a color printer has 7 basic pigments; CMYK + Orange, (Mint) Green + (Violet) Blue. Nature has way more. Nature also has odd ways to create color for the camera or human eye and the color impressions in both may not be the same. No target chart that covers all possibilities.
Cut silkscreen, car paints and/or artist's acrylic sample books to patches and glue them on OBA free white museum board. Good chance you can increase the pigment variety. Spectral plots can tell you which patches have unique pigments. Study Ben Goren's approach and ArgyllCMS features to make that work. A fresh spectrometer is fundamental too. Old plan, never done the hard work to finish it.
As I understand it the dual lighting profiles approach is worth more than the sum of two profiles, in practice.
--
Met vriendelijke groet, Ernst
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
July 2013, 500+ inkjet media white spectral plots.
-
Hello Ernst Dinkla and thank you for your answer!
Yes, as I said I will add many many patches of various material so to have a very wide range of colors in addition to the patchs that I will print.
My idea was just about Goren's one (he printed some patches as well...) :-)!
I did not get when you say As I understand it the dual lighting profiles approach is worth more than the sum of two profiles, in practice.
...
Do you mean that DNG profiles are better than ICC or what? Sorry I really did not understand...
Unfortunately I don't own a spectrophotometer but I hope a colorimeter would do as well...
Thanks a lot!
Ale
-
Yes, as I said I will add many many patches of various material so to have a very wide range of colors in addition to the patchs that I will print.
Keep in mind that your camera's Raw gamut capture capabilities will be limited by the display you view color on. Hopefully you'll have a display that's close to AdobeRGB.
Digital sensors capture voltage readings from each charged pixel cell according to how many RGB photons are filtered to each by the RGB Bayer filters. It's all rendered as grayscale and reconstructed by software and demosaicing algorithms and assigned color using a color managed and calibrated/profiled display.
Another consideration is there are currently modern inkjet printers whose inks go beyond AdobeRGB gamut. (See 3D Gamut plot comparison below). I found this out recently printing a cyan colored dress on a Fuji Frontier DL430 inkjet on Fuji Dry Lab Glossy paper whose cyan my camera could not capture and my display could not reproduce. But I still can't be sure which device failed at capturing and reproducing this cyan because I'm using an sRGB gamut display to view the Raw image and a custom DNG profile.
This is not to say you won't be able to capture and build a profile of most of what can be photographed in artwork and paint media. I'm just relating the limitations involved in all the possible color combinations your camera/display can capture and display. You still may come across some type of paint, ink or combination of fixative in a piece of artwork that fluoresces beyond what these devices can capture and reproduce. This may be fixed with HSL adjusts and maybe not.
-
Keep in mind that your camera's Raw gamut capture capabilities will be limited by the display you view color on.
I hope that's a typing error, because it's wrong and misleading. The capture capabilities of a sensor are not at all controlled by the display's capabilities.
I suspect you meant to say that you won't be seeing the full gamut of an image unless you have a very good monitor. Even then it's highly probable that there will be colours that won't be correctly displayed and clipped by the monitor.
It's certainly more important to get a good wide gamut monitor that's calibrated and profiled correctly and understand it's gamut limitations. Then maybe get some good software that allows you see what happening with gamut comparisons before worrying about the esoterica of custom profiling sensors.
-
I hope that's a typing error, because it's wrong and misleading. The capture capabilities of a sensor are not at all controlled by the display's capabilities.
I suspect you meant to say that you won't be seeing the full gamut of an image unless you have a very good monitor. Even then it's highly probable that there will be colours that won't be correctly displayed and clipped by the monitor.
It's certainly more important to get a good wide gamut monitor that's calibrated and profiled correctly and understand it's gamut limitations. Then maybe get some good software that allows you see what happening with gamut comparisons before worrying about the esoterica of custom profiling sensors.
You're not an authority on determining whether something is misleading on this topic because so few have made a connection to color reproduction theory that's functional for photographers and repro experts. I stand by what I said because no one in the digital imaging industry is going to see a darn thing with regard to color unless they can determine what each of the sensor pixel sites recorded. And since your regular consumer grade digital cameras are not a spectroradiometer, we're all taking the best educated guess on what's going on within the electronics of all digital capture and repro devices. Until some one can prove visually all the colors that are capable of being reproduced by such devices I'm sticking to my understanding on how all this works and stand by what I said.
Grayscale is what is handed to the A/D and converted to 1's and 0's and is then redefined by software and a display device on a computer with regards to all the colors that are possibly reproduced. It still does not prove all the colors possibly captured.
If you have proof that isn't some complicated black box theoretic driven graphic but is functional for photographers and repro experts that shows to the human eye all the colors a camera can reproduce and be seen on a display then show us the proof.
First order off the bat, I want you to capture with your camera the cyan from that Fuji DL430 print I mentioned. Post it here and those with AdobeRGB gamut displays will tell me if they are seeing a color they've never seen before in the history of color repro. I'm pretty sure you're going to turn me down.
-
It still does not prove all the colors possibly captured.
I guess the issue is with the word "capture", if because of CFA properties a lot of different colors can't be really distinguished after the "capture", that does not mean they were not "captured"... you have the charge in your sensels after exposure and it's not noise, so what what is the word you 'd use to describe that if not "capture" ?
-
You're not an authority on determining whether something is misleading ..... I stand by what I said........
I may not be an "authority", whatever that means, but your original sentence is just wrong. Go back and read what you've written in comment #8
"your camera's Raw gamut capture capabilities will be limited by the display you view color on"
Do you really believe that ?
If so, how does it work ?
If you took your sentence at face value the camera would need to know what display is being used for the image to save the image's data values. Whenever I've shot RAW my camera it has never needed to know what display I'm using to know what data to save. Since capture that data hasn't changed as I've changed the display I use.
By making such obviously incorrect statements you discredit yourself for any further observations, worthwhile or not.
-
Hi Tim
A camera can capture all visible wavelengths of light and beyond! Try using a monochromator at 5mn intervals ranging from 360 to 730 nm. You get pretty good results.
I think what you are trying to say is that the 'reproduction' of the camera raw capture is limited by the current display/printer/projector technologies and their associated gamuts.
Iain
-
I guess the issue is with the word "capture", if because of CFA properties a lot of different colors can't be really distinguished after the "capture", that does not mean they were not "captured"... you have the charge in your sensels after exposure and it's not noise, so what what is the word you 'd use to describe that if not "capture" ?
We might as well have a metaphysics discussion on this topic because you can't know what you can't see and the only way anyone is going to see the gamut capture capabilities of a digital camera is through the limited gamut of a display.
I'll accept your interpretation of the concept of the gamut capture capabilities of a digital camera. What I won't accept is anyone calling it a fact without proving it as a functional concept for photographers and repro experts to use out in the field. That hasn't happened, yet.
But I have hopes that the technology and the accompanied theories behind how it works will be made more clear for us non-scientific regular photographer/repro experts. My discovery of this unreproducible cyan was my real world connection to what these 3D gamut plots and theory is trying to convey in a way I can understand.
Now on to proving the other unimaginable colors that can or can't be reproduced by a digital imaging system from capture to display to print.
-
"your camera's Raw gamut capture capabilities will be limited by the display you view color on"
Do you really believe that ?
From a standpoint on whether you can prove it, yes.
If a digital camera can capture a huge gamut in its own Raw format and there's no one there to see it, does it really exist and if you can't see it, is it really useful to the photographer? I say NO, it isn't useful.
I say prove it so it's a functional concept.
-
Hi Tim
The camera can capture all (open for debate) colours we can see, the problem is we cant reproduce them all (yet).
-
Hi Tim
A camera can capture all visible wavelengths of light and beyond! Try using a monochromator at 5mn intervals ranging from 360 to 730 nm. You get pretty good results.
I think what you are trying to say is that the 'reproduction' of the camera raw capture is limited by the current display/printer/projector technologies and their associated gamuts.
Iain
I'll accept that explanation, Ian. But it still doesn't prove all the possible colors a digital camera can capture and be represented in the form of 1's and 0's without a display device.
Recorded wavelengths do not a picture make especially when a human gets a hold of that data in an editing session in order to reproduce what they saw at the time of capture which can be quite subjective due to human adaptation which a machine hasn't a clue about.
For instance I'm puzzled by the fact I see digital captures of intense orange and yellow dayglow colored utility vests worn by street and construction crews that look quite accurate according to how I recall their level of intensity. When I sample the actual RGB numbers, there not as wide gamut as I'ld thought mainly because I'm viewing these intense colors on an sRGB gamut display.
Human optics and perception is the sand in the vaseline to machines that strictly define gamut as recorded in the form of wavelengths when it comes to accurately reproducing the gamut capture capabilities of a digital camera and imaging system.
-
Hi Tim
The camera can capture all (open for debate) colours we can see, the problem is we cant reproduce them all (yet).
note about metameric failures, you can can capture a pair of certain different wavelengths, but you will not be able to distinguish them afterwards based on the data obtained from the camera afterwords - hence Tom (and his software) does not know whether it was really a color A or a color B, even if you can reproduce them both...
-
Hello and thanks for all the interesting responses!
I think this thread is going a bit off topic... ;D
Going back to the original DIY charts...
I see the main point that many of you raised is that, simply printing a chart wouldn't be enough for an accurate profile and I see the reasons for it.
After a bit of thinking I came up with this idea (that obviously is not entirely mine): why not to use more, completely different, charts to create just one profile?
Argyll enables you to read various chart and combine all the data in a profile, therefore why not to use (just to make an example but please let me know your suggestions):
- Chart 1: IT8
- Chart 2: ColorChecker
- Chart 3: DIY Printed Chart (up to you what colors to use here but I would use the "cardinal point" of the LAB and then make gradation)
- Chart 4: DIY Acrylic Color Chart (what do you suggest to use as color?)
- Chart 5: DIY Filter Color Chart (create a frame with many pieces of various filter, ex Lee Filter Swatch Book, and then retroilluminate everything)
- Chart 6: A detailed greyscale + light trap + super bright reflective material for white.
I know it might be a super long process but I think in this way you could cover many more colors than a normal commercial chart...
Looking forward to hear your suggestion!
Thanks a lot!
Ale
-
Hello and thanks for all the interesting responses!
I think this thread is going a bit off topic... ;D
Going back to the original DIY charts...
I see the main point that many of you raised is that, simply printing a chart wouldn't be enough for an accurate profile and I see the reasons for it.
After a bit of thinking I came up with this idea (that obviously is not entirely mine): why not to use more, completely different, charts to create just one profile?
Argyll enables you to read various chart and combine all the data in a profile, therefore why not to use (just to make an example but please let me know your suggestions):
- Chart 1: IT8
- Chart 2: ColorChecker
- Chart 3: DIY Printed Chart (up to you what colors to use here but I would use the "cardinal point" of the LAB and then make gradation)
- Chart 4: DIY Acrylic Color Chart (what do you suggest to use as color?)
- Chart 5: DIY Filter Color Chart (create a frame with many pieces of various filter, ex Lee Filter Swatch Book, and then retroilluminate everything)
- Chart 6: A detailed greyscale + light trap + super bright reflective material for white.
I know it might be a super long process but I think in this way you could cover many more colors than a normal commercial chart...
Looking forward to hear your suggestion!
Thanks a lot!
Ale
Sounds like a good idea, Ale. Give it try and see what happens and report back.
I mean from your response you seem to be knowledgable enough to detect or sense that laborious process would yield better results, so I'm a bit curious to know what gives you that idea.
With regards to perception of accurate reproduction of color this page on color perception might present some concerns...
http://colorindesign.net/2010/10/12/the-changing-nature-of-color/
Keep in mind that even though perception of color will change, a machine (spectrophotometer) will still define it as the same. IOW you may be working to hard on this due to these perception issues.
-
Hello everybody...
I am going forward and so I am preparing all the various chart that I need to experiment this process.
I am done with the layout in photoshop and I have prepared the *.cht files to make argyll able to recognize the charts when using "scanin".
Still I am facing some problems and so I thought of dropping a line here to see if anybody could help me...
In order to run the command "scanin" and getting that *.ti3 file needed to create the ICC profile with the command "colprof" I need to have a *.cie file containing the L*a*b* values (to be acquired with the spectrophotometer).
In Argyll online documentation there are no specification about how this file should be formatted (I can guess it by looking at other charts *.cie files), is there anyplace online where I should look for this information?
Also, which command do you suggest to use (in Argyll) to create this *.cie file? As I guess I can't make it directly I thought of using "chartread" to create a *.ti3 file and then convert it to *.cie with "spec2cie". But in this case I would need a *.ti2 file and, once again, there are no specification about how to make this file...
Thanks a lot for your advices!