Profile Assessment

[Emulator]

In a profile Round Trip assessment am I right in assuming that Maximum Delta E should be less than integer value 6, to acceptable?

I seem to be finding some profiles with much higher values, (20 to 50 in some rendering intents),
and in one case 90 in a profile that produced very poor results.

PRMG Interoperability Round Trip Results seem to expect a wider range of values of DE up to 10.

Any guidance would be appreciated.

 

[Emulator]

The round trip test as I understand it, tests the AtoB Tag input tables (used for example in soft proofing) and the BtoA Tag output tables (used by the printer) as a round trip and shows the combined result ( i.e. the resulting characteristic should match the input). The tests are applied to each of the three intents.

There may be defective values in either of the input or output tables, both have to be within limits to be acceptable (ie below 6 combined to meet the 2000 spec.).

The image (click to enlarge, or right click and "open link" to enlarge print size) shows the results of a round trip test on a printer profile.


This shows that the three profile intents, relative colorimetric, perceptual and saturation have Max DeltaE values below the desired maximum limit of 6.

Note added later:
An excellent Xrite PDF explaining colour management, terms and definitions, can be found at:-
http://www.xrite.com/documents/literature/en/L10-001_Understand_Color_en.pdf
The definition of DeltaE is shown on page 14.

 

[Emulator]

It is surprisingly difficult to find readable and understandable (without hours of study) guidance on colour management. So as I find references which provide good reading on the subject, I will provide the links on this series of posts. One here on LUTs, is I think a good one.
http://www.lightillusion.com/luts.html

If you want to see the detail in profiles down load ICC Profile Inspector (to see the tag structures) and wxProfileDump (to do the round trip tests), both free on the ICC website.

 

[Grandad35]

The round trip test as I understand it, tests the AtoB Tag input tables (used for example in soft proofing) and the BtoA Tag output tables (used by the printer) as a round trip and shows the combined result ( i.e. the resulting characteristic should match the input). The tests are applied to each of the three intents.

I'm not sure that it is a given that the round trip should give back the original image.

The purpose of a soft-proof is to show what the printed image will look like after it is printed. Assume that a printer has a very poor capability to print reds, and that all of the printed reds are desaturated as a result. The soft-proof of this image should also be desaturated in the reds, and won't be anywhere close to the original image, giving a large DE from the original.

Test a copy of the same image after lowering the contrast to push the histogram away from the edges (e.g. use Photoshop's "Levels" adjustment with the "Output levels" changed from 1-255 to 60-195). This should improve the round trip DE by eliminating colors that are out of (or even close to) the printer's gamut. Conversely, a highly saturated image should make the round trip results worse.

 

[Emulator]

My comments relate to AtoB and BtoA tags, not images. AtoB transforms should give the inverse of BtoA tranforms, thus can be used in softproofing an output profile. If they don't match the inverse, the display derived from use in comparison will not correctly match a printer profile output. The mismatch might then be blamed on other factors and result in incorrectly adjusted images. The Max DeltaE values are obtained by running a round trip in ICC wxProfileDump.

 

[Grandad35]

My comments relate to AtoB and BtoA tags, not images. AtoB transforms should give the inverse of BtoA tranforms, thus can be used in softproofing an output profile. If they don't match the inverse, the display derived from use in comparison will not correctly match a printer profile output.

As I understand it, for an “output” profile:

1. The BtoA LUTs are used to transform the R/G/B color values in an image to the R/G/B color values that are sent to the printer (the rendering intent determines which of 3 the LUTs is used).

2. The AtoB LUTs are used to transform the R/G/B color values that are sent to the printer back to the image’s (typically sRGB) color space for soft-proofing (the rendering intent determines which of the 3 LUTs is used). It must then be further transformed by the BtoA LUT in your monitor profile before being sent to your calibrated monitor to get an accurate soft-proof.

3. As these 3 dimensional matrices are just tables of numbers, they have no real meaning except in how they transform the R/G/B color values for each pixel in an image.

If the AtoB and BtoA LUTs were the true mathematical inverse of each other (giving a round trip DE of 0), then wouldn't the soft-proof of an image always be the same as the original? If this were the case, how could soft-proofing have any value?

Do you have a link that indicates that a round trip should return the original colors for a good profile? This link on Evaluating an output profile speaks of using the round trip technique using the colorimetric intent, but only as a method to keep all of the color values within the gamut of the output device (printer) before the final comparison with the measured colors on the printed image. If there are out-of-gamut colors in the original image, the round trip will obviously result in changes to the color values in the out-of-gamut areas.

More reading:

What is an output profile?
What happens on printing
Profile tag definitions – see Table 25 in section 9.2

 

[Emulator]

Hi Grandad35,

This is a lovely subject for discussion, it could go on for years!

Incidentally the question of max deltaE tolerance in practice, is I think 6, after numerous web searches, typically http://w3.efi.com/en/services/fiery-wide-format-services/~/media/Files/EFI/COM/Services/Delta E_H_T.pdf


The best ICC document with clear descriptions of profile content that I have found is: http://www.color.org/ICC_Minor_Revision_for_Web.pdf

Pages 18 and 19 provide the data on input and output tags.

In your para 1 you refer to transforming values in an "image", the tag descriptions prefer to relate to transforms, from or to, Profile Connection Space (PCS) and input and output devices. I think this is because the application of a profile can be so varied.

It seems that creating the true inverse of the input or output tag is not easily achieved in practice, because the issue is extensively raised in discussion and visually discerned in practice. If you look at the delta E values in profiles in practical use, the range of values varies widely. Using ICC wxProfileDump, Round Trip Report, provides numerical values.

A useful description of the round trip process is at:
http://www.gamutvision.com/docs/roundtrip.html

If the inverse transforms were perfect and the display correctly profiled, the softproofing process would illustrate a true representation of the output to the printer. Not necessarily the printed image, as this would depend on how good the profile was in the first place.

Regards for now,
Ian

 

[Grandad35]

In your para 1 you refer to transforming values in an "image", the tag descriptions prefer to relate to transforms, from or to, Profile Color Space (PCS) and input and output devices. I think this is because the application of a profile can be so varied.

I think that we are describing the same thing, but using different words. In my understanding, the transforms to/from the PCS and other devices are actually the transforms of the color values assigned to each pixel.

If the inverse transforms were perfect and the display correctly profiled, the softproofing process would illustrate a true representation of the output to the printer. Not necessarily the printed image, as this would depend on how good the profile was in the first place.

Here, we disagree. As is stated in this link on soft-proofing:

Soft proofing is simply a mechanism that allows you to view on your computer monitor what your print will look like when it is on paper. A specific paper. That paper and ink combination has been defined by the profile that you or someone else has made for your printer / paper and ink combination.​

The soft proof does not show the data that is sent to the printer (if it did, the AtoB and BtoA LUTs would be the inverse of each other), but what the finished printed image will actually look like after it is printed. This includes handling the rendering intent, any out-of-gamut colors, non-linearities and color casts for the ink/paper/printer used to generate the profile and print the image.

 

[Emulator]

In your last para, I think we are talking of slightly different things. Softproofing is applied to the profiles as well as images, the purpose being to detect the differences between the BtoA and AtoB tags, which should ideally be perfect inverses, if the softproofing of images is to be accurate. See the Gamutvision ref. in post #7

pharamcist kindly generated an Argyll 1.6.1 profile for Sihl glossy 280g for my Canon Pro9000 II on IS inks, this determined the profile tags enumeration. The next question is how good are they? The images below are generated in Gamutvision to illustrate the results of a "round trip" of the profile. In the first image, the input is the net, output is the solid. Here we are looking for gaps between the net and the solid. The differences are very small, which is good. The Gamutvision shows the gamut volume as 787922 L*a*b* volume units (Delta E^3).

The second image shows the delta E differences, between the input and output tags. Notice that all are comfortably inside the suggested pass level of 6 (detectable difference).

 

[Emulator]

I came across this description of soft proofing in a luminous-landscape.com article.
Soft Proofing
Soft proofing is simply a mechanism that allows you to view on your computer monitor what your print will look like when it is on paper. A specific paper. That paper and ink combination has been defined by the profile that you or someone else has made for your printer / paper and ink combination. When a printer profile is made the colour of the paper is one of the factors that is figured into the profile, because the spectrophotometer is reading the combination of the ink, and the paper that lies beneath it.
So, if you were able to view your image through the printer profile, you would be able to see how that particular combination of ink and paper would reproduce it, taking into account the gamut as well as other characteristics of the inks used.