Using ArgyllCMS + Colormunki to produce excellent printer profiles

[nrdlnd]

Yes, they are very similar. I cannot see any significant difference in Gamutvision.

This method, although only a simulation, is more sensitive than comparing two prints side by side where differences of a fwe delatE may not be noticed. And of course it is very quick and very low cost!

I think this way: Of course it's interesting to be able to toogle between different profiles in a softproof (and see if there is any difference). But if I have analyzed a profile in Gamutvision and I can see that this is a good profile. Then I can use it. I can use softproofing as a help. I think it's still not possible to rely on the softproof. The screen may be in another color space and then when I softproof using the profile for a matte paper it will look soft and without contrast. The print can give quite another impression. It can look more contrasty (more 'graphic') as it has a more limited dynamic range. Most of the time I want to print my pictures on papers that give a wide dynamic range (like Canson Baryta Photographique or Canson Platinum Fibre Rag). The pictures I'm working with now I want to give another impression, and then the softproof will only give an indication of what the result will be.

Per

 

[RogerB]

I think this way: Of course it's interesting to be able to toogle between different profiles in a softproof (and see if there is any difference). But if I have analyzed a profile in Gamutvision and I can see that this is a good profile. Then I can use it. I can use softproofing as a help. I think it's still not possible to rely on the softproof. The screen may be in another color space and then when I softproof using the profile for a matte paper it will look soft and without contrast. The print can give quite another impression. It can look more contrasty (more 'graphic') as it has a more limited dynamic range. Most of the time I want to print my pictures on papers that give a wide dynamic range (like Canson Baryta Photographique or Canson Platinum Fibre Rag). The pictures I'm working with now I want to give another impression, and then the softproof will only give an indication of what the result will be.

Per

What I was suggesting was a method to compare the rendering of two profiles directly, not using soft-proofing to simulate the printed output. The comparison does not rely on the accuracy of the monitor or any other parameter. It's a way to compare two profiles without using a program like Gamutvision. Of course, if you have Gamutvision you can do the same thing - like this.

 

[Emulator]

I hadn't realised that the gamut of the Epson 3880 using pigment inks, was so much larger than typical dye ink printers, until I saw your results in posts# 101 &103!!

I like your error display in #112, hadn't thought of that. Stops all arguments!??

 

[RogerB]

I hadn't realised that the gamut of the Epson 3880 using pigment inks, was so much larger than typical dye ink printers, until I saw your results in posts# 102 &104!!

I like your error display in #113, hadn't thought of that. Stops all arguments!??

It seems that almost everyone believes that dye inks give a wider coour gamut than pigment. Without wishing to start a gamut war, I believe this is a bit of an urban myth. Of course a lot depends on the paper, and the baryta used by @nrdlnd is ideally suited to pigment ink. Dye inks do not generally perform well on baryta papers in my experience, being better suited to RC types. Even so, my 3880 gives virtually the same colour gamut (918,035) on a reasonable microporous RC paper. It certainly has a different "look" from a dye ink print, so each to his own I guess.

 

[nrdlnd]

It's a way to compare two profiles without using a program like Gamutvision. Of course, if you have Gamutvision you can do the same thing - like this.

Can you please tell me what this picture shows and the settings you have chosen!? I don't understand how to interpret it. Probably I will not strive to be a guru of the Gamutvision program but it seems to be a valuable tool for evaluating profiles. I have searched the forum but I can't find a 'tutorial' on how to use it and its possibilities.

Per

 

[Emulator]

There is extensive help/tutorials on GamutVision, click on "help" on the GamutVision working screen or go to the website. I found it was useful to print all the many pages and put them in a file, for reference.

 

[RogerB]

Can you please tell me what this picture shows and the settings you have chosen!? I don't understand how to interpret it. Probably I will not strive to be a guru of the Gamutvision program but it seems to be a valuable tool for evaluating profiles. I have searched the forum but I can't find a 'tutorial' on how to use it and its possibilities.

Per

Sorry if it is confusing. This is my explanation of what is happening;-

We load the Grainger rainbow image into Gamutvision using the Read Image for analysys and set the input profile (1) to sRGB. This limits the actual colours to something closer to the printer gamut. You can of course use any image for this purpose.

The profile (2) is one of the profiles that we want to compare - in this case your 1444 profile.

Gamutvision converts the image to the colour space defined by the profile. You can view the result like this. The scale shows the colour errors in the printed version of the Grainger rainbow using your 1444 profile.

The output from (2) contains the RGB values that would be sent to the printer when printing this image. We send this set of RGB values to (3), together with the profile from (2). We then use the profile in (4) to calculate the colours that result from sending the RGB values directly to the printer using profile (4) - setting Rendereing to None is printing with No Colour Management.

The colours calculated from (4) are compared with the colours calculated in (2) and the difference is displayed. In this cse the given RGB values will produce colours using the 1156 profile that are all within 1.1 deltaE of the colours produced by the 1444 profile. In other words the profiles are essentially identical.

Here's what happens if I use a different profile in (4). In this case I have used one of my profiles for a matte paper. You can see that the response in the green region is very different, so these profiles are not interchangeable.

I hope that helps to clarify things.

 

[nrdlnd]

I hope that helps to clarify things.

Thank you very much for that clarification! I can see that deltaE is 1.1 and understand that mathematically this is very similar. What I don't understand is what I can see on the screen. What does the different colors and the pattern mean? When I see the view of your profile of a matte paper I can see that the colors are different and you say that it differs especially in the greens? How can I see that?

Then the next question: When I make a preprofile I can see in the next profile with this profile embedded a great change of the look of the chart. I can see a lot more green patches especially. It seems that Argyll needs a lot more patches in the green area to make a better profile. Why is it like that? Is the green the most difficult to get right? I think I'm missing some elementary understanding of colors.

Per

 

[RogerB]

Thank you very much for that clarification! I can see that deltaE is 1.1 and understand that mathematically this is very similar. What I don't understand is what I can see on the screen. What does the different colors and the pattern mean? When I see the view of your profile of a matte paper I can see that the colors are different and you say that it differs especially in the greens? How can I see that?Per

Sorry - when one is familar with an application it is easy to forget that others may not be!

The large pane in the display represents the image that has been read in to analyse the profile performance. This image is also shown in the small pane at the top right. The large pane can display many different functions but I have chosen to display the colour errors. You need to visualise the test image underneath this error display. The errors are displayed as a colour map, where each colour represents a number as shown in the scale below the pane. For my profile the largest errors (darkest colours in the colour map) are in the region of the test image that is mainly green.

Then the next question: When I make a preprofile I can see in the next profile with this profile embedded a great change of the look of the chart. I can see a lot more green patches especially. It seems that Argyll needs a lot more patches in the green area to make a better profile. Why is it like that? Is the green the most difficult to get right? I think I'm missing some elementary understanding of colors.
Per

I am not an expert on Argyll CMS, but I believe that using a preprofile allows targen to place the test points (colours) so that their distribution is nearer to the actual printer gamut rather than in a simple cubic matrix. Since inkjet printers have good response in the green (look at the shape of the 3D gamut plot) this may well result in a bias towards the greenish patches.

 

[Emulator]

I can see that deltaE is 1.1 Per

Notice that the scale has a maximum of 1.1.
Against the scale is shown a coloured scale. Each colour represents an associated number in the range 0 to 1.1.

The main coloured display, above the scale, represents the images (modified by the profiles under examination) and shows the errors between the two images being compared. Read the error values by comparing them to the scale colour values.
You then have an indication of the Delta E error between the two image profiles at any point in the image.

The output of the image modified by the profile is displayed top right. Roger's two arrows show the related points on the image, the profile Delta E display and the value scale, for a greatest difference of green, between your and his profiles.