The rapid absorption does have the disadvantage that the colours become desaturated as the inks move deeper into the paper. What the printer sees as he pulls a fresh copy from the press for checking the inking is notably more colourful than that which the reader sees 4-8 hours afterwards. Generally, this doesn’t present a problem as the advertising customer and the reader see the printed result at a time when the inks have penetrated the paper and the absorption rate has slowed to insignificant levels. The ICC colour management profile, which is used to make the picture colour separations for the advertisement proofs, is itself made from the measurements of printed profiling test charts that are more than 8 hours old. For the ad customers and readers, the printed result is exactly as expected.

However, this quite noticeable difference in the image appearance as seen by the printer and the same copy seen some hours later by the advertiser is important. Obviously, the “dryback” effect is most challenging for the pressman who has to print the advertisement in a way that satisfies the customer even when viewed much later in time. Printing quality colour in newspapers is somewhat akin to hitting a moving target, but fortunately, colour management does provide a solution.

Some researchers have documented the “dryback” on newsprint, but without mentioning the practical implications with regard to newspaper printing and soft proofing. The surge of interest in monitor soft proofing has meant that perhaps a different soft proof image with more saturated colours – a “strong colour” version of the image – is needed by the printer as an accurate guide. Judging the levels of inking needed against a more vivid colour image would result in a better colour match at the press and one that eventually corresponded better to the less colourful but reproducible colours promised by the contract proof originating out of the standard ICC newspaper profile e.g. the ISOnewspaper26v4 profile.

How different are the images presented to the printer and to the reader?

Fig. 1 The newsinks on newsprint gamut volume shrinks rapidly after printing. Here, the colour range is reduced by more than 11% after just 30 minutes.

Fig. 2 Top and bottom views of the shrinking newsink-newsprint gamut after 2 minutes, 4 hours and 24 hours. After 4 hours, the gamut has lost almost 15% of its colour.

The grey areas show the gamut boundary of the ICC ISOnewspaper26v4 profile. The freshly printed colour gamut retracts to a size even smaller than the de facto standard newsprint colour gamut in this example.


Fig. 3 A collection of 22 freshly printed colours that changed more than 4.5 CIELAB Delta E (1976) units within a period of 4 hours. The colours in the graph are arbitrary colours and not those of the measured colour patches. In reality, they are mostly dark 3- and 4-colour halftone combinations.

Fig. 4 Cardinal colour changes due to “dryback”. The changes are normally more for the 2-colour R,G,B overprints than for the primary CMYK inks.

Fig. 5 To allow for the “dryback”, the soft proof monitor image which the printer uses for colour matching should be more colourful and with deeper shadow tones than the hardcopy sent to the advertiser as a contractual proof.

The measurement data was taken from a CMYK test chart for producing print characterisation data for an ICC soft proof monitor profile, the basICColor CMYKick chart.

All the data are averages of the same four samples measured at the indicated intervals in time.

In trying to improve the match between the freshly printed image and the soft proof, it’s important to note:

1. The on-screen RGB differences between the 2 minute representation and the 4 hours representation are quite noticeable, but there was no real difference at all between the 4 hour and the 53 hour representation.

2. The differences in appearance between the freshly printed copy and the same copy after 4 hours or more, affects all tones - dark, middle and light tones.

3. The change in colour is not the same for all colours. The dark tones, browns and dark neutrals show the most difference. Two-, three- and four-colour overprints show more change than the single colours. Light colours are less affected.

4. Even though the measured CIELAB Delta E (1976) differences were greatest and most noticeable in the dark tones, notable differences were visible in the skin tones, although the calculated Delta E values were much smaller. European skin tones often have relatively light halftone- and ink values, so that large Delta E differences are not likely to occur but the change is relatively proportionate to the percentage of Total Area Coverage (TAC) of the halftone combinations.
In any case, the human visual system is more sensitive to changes in skin tones and pale colours with low chromaticity than to the same change in dark saturated colours.

5. There is a strong statistical correlation between the TAC values of the colour patch and CIELAB Delta E values. The CMYK test chart contained 336 colour patches, with TAC values ranging between 0-400%. The calculations included monitoring the colour difference for each patch during the “dryback” period. The colour patch TAC values and their final measured colour difference Delta E values were positively correlated (0.851) i.e. high TAC values are strongly related to a high colour change.

6. Like all CMYK profiling test charts, the chart has no Grey Component Replacement (GCR) in the halftones. That function is part of subsequent processing by the ICC profile-making software in which the lightness value of the CMY components of a colour can be partially replaced by an equivalent amount of black ink. The TAC values are substantially reduced by GCR and meaningful quantities of costly CMY colour inks are replaced with the less expensive black ink.

In practice, as CMYK test charts are not GCR processed images, data and observations regarding colours with TAC values between 241 - 400% are unusable. Nevertheless, even investigating the 287 colour patches with TAC values of 240% or less, the coefficient of correlation statistic is still 0.805, which also strongly suggests that high TAC values lead to very significant CIELAB Delta E colour changes during “dryback”.

7. From Fig. 4, the black ink shows considerably less colour change than the other primary colours of C, M and Y. This may be a typical result - see also IFRA Special Report 2.39.

There are three points to note here:
a) Ink sequence and the amount of additional ink applied to an already printed area play no role here because the inks are not overprints in this case.
b) The measured primaries are not halftones, but solid printed areas.
c) Single full-tone primary colours are rare occurrences in printed real world images. These colour patches are less of a concern than the behaviour of the multi-colour halftones.

Even though the black ink often has the highest dot gain value (29% here) and potentially a higher rate of absorption than the other colours, the full-tone black is a relatively minor contributor to colour change during dryback. This complements the observation that a low chroma monochrome colour doesn’t or can’t register high CIELAB Delta E values, normally.

In calculating colour difference using the CIELAB 1976 formula, Delta-E, CIELAB a* and b* are the two colour elements of the three elements considered – the third element being L*, lightness. If these colour elements, a* and b*, must necessarily have very low numeric values, which certainly applies in the case of black as the least colourful of the 4-colour process inks, then black inks don’t and, more importantly, can’t normally introduce or be the cause of large Delta-E colour difference values.

This is also one reason why GCR works so well in stabilising newspaper four-colour printing, when part of the multi-colour colour tone is replaced with an equivalent single-colour black halftone.

8. CIELAB values are not colour appearance values. The difference in appearance is rather more marked than the CIELAB Delta E figures seem to indicate.
The reasons may be related to surface ink film thickness changes, lighting from adjacent surfaces (flare) or even some small degree of gloss or texture changes – influences that are included in visual assessments but not included in instrumental colour measurement.

9. Gamut volume changed by 11% in the first 30 minutes, by 14.5% within the first 4 hours and by 16% after 53 hours.

10. Tone Value Increase, TVI, (dot gain) for a 50% original tone, didn’t change significantly over the measured time period, according to X-Rite ProfileMaker’s spectral data calculations – cyan 26.8 ± 0.3%, magenta 20.1 ± 0.3%, yellow 20.3% ± 0.6%, black 29.1% ± 0.1%.

11. In the CIELAB system, the L* linear lightness scale matches the human visual response quite well, however, the C* chroma scale does not. The mismatch between calculated and perceived colour differences is worse at higher chroma values. The DE2000 colour difference formula is better than CIELAB Delta E at describing the visible differences when chroma values are low.

12. There was no obvious sign of the “blue-changes-to-purple on drying” problem that has sometimes been reported. In the past, the problem has been variously diagnosed as a colour separation problem or an ink problem. The test chart is a CMYK colour-separated file, so there were no colour deviations due to colour separation remapping e.g. gamut compression.

13. A “strong colour” soft proof profile is not the only way to simulate higher colour contrast, but it’s the most flexible, controllable and appropriate method.
The ease with which an ICC soft proof profile can be inserted at an appropriate point in the soft proofing workflow will depend on the system.

14. The “strong colour” soft proof profile is a distinct improvement when monitoring the production of newspapers printed using standard news inks on newsprint. However, such a soft proof profile only needs to be in the monitor RGB workflow when using these materials. With higher quality printing materials, a false impression will be given by using a “strong colour” profile. In commercial heatset web offset, the ink absorption is not as dramatic and the ink film stays more on the substrate surface where it remains, set by the high speed driers. Colour changes are not a problem. Further, it means that for heatset offset, the same ICC profile which is used for the image colour separation can be used for soft proofing.

15. The ideal soft proof viewing conditions are difficult to establish and maintain. In the press control room, the lighting and reflections from coloured surfaces combined with the difference between the self-illuminating monitor display and the printed copy influence the comparison between a soft proof and the printed result. Nevertheless, effective soft proofing is possible, if suitable viewing conditions are provided and the imaging parameters carefully controlled.

16. An ICC profile derived from the CMYK ink and paper characteristics measured soon after printing is not enough for a soft proof display. Depending on where you take your soft proof image data, dot gain may- or may not be included.

The standard CMYK profile-making software normally provides the correct CMYK halftone dot sizes to be imaged on the plate, without the dot size growth that is anticipated from dot gain on the press. It’s generated from measuring a standard CMYK profiling testforme e.g. the ISO 12642 or ANSI IT8.7/4 and the dot gain data is calculable as the original dot sizes are known. It’s also then possible to generate a soft copy print version of the CMYK image on the screen from either RGB or CMYK data by using the colour separation program e.g. Photoshop. However, if you're in pre-press, you're nearly always working from the standard CMYK profile, and, of course, errors further in the workflow aren't checkable and catchable, unless you are using the very same RIP data and settings as the production people. A true soft proof isn’t quite possible. It's almost WYSIWYG but halftone screening and data errors can creep in between the pre-press workstation and the RIP output. Also, the page and imposition data is not usually available to make a page soft proof suitable for the pressman.

The alternative is a soft proofing system that works from TIFF/G4 1-bit data - the same data that is used to image the plate. There are several of these. The claimed advantage for using TIFF/G4 production RIP data is that any late errors in the text & images that arise accidentally, imaging system calibration errors or screening effects, as well as plate colour composites checks to prevent colour swaps, and the addition of colour control- and register marks are in the soft proof. Hopefully, all possible errors can be corrected before being printed or caught very early in the print run. Many workflows are now "late binding" operations, where the image servers do the separations automatically, just before passing the complete page data to the RIP and CTP. Using TIFF/G4 data allows the building of a WYSIWYG operation. BUT the production TIFF/G4 data doesn't have the dot gain info and a soft proofing system centred on this type of data has more to do to simulate a "strong colour" proof, though it's using this real-time data in an enhanced colour proof that makes most sense for the printer.

A soft proof needs to simulate the printed result and so must display colours that include the dot gain expected from the press. Once the dot gain characteristics are calculated from the soft proof CMYK profile measurement data, it’s then possible to also make an initial “strong colour” soft proof profile that can complement the software that converts TIFF/G4 1-bit colour-separated data to 8-bit full-colour composite image data.

A more vivid colour soft proof profile is not the only deciding factor in simulating the fresh print, but it is one significant step in that direction. The monitor characteristics, the viewing environment, ink, paper and overprint ink sequence are all important considerations. There are also a great number of other factors to consider in choosing a soft proofing system in which the display matches the print product at the press and by the advertiser. These will be the subject of a WAN-IFRA Special Report that is in preparation. In the meantime, if any member has a comment on this study or needs help with soft proof profiling, or soft proofing generally, we’ll try and assist.