uV Cut D50 (M2) or full uV D50 (M1)? Where it makes a difference

[Doug Gray]

Here I discuss why M1 v M2 is critical for graphic arts and accurate hard proofing as well as why it makes almost no difference for photo printers. I include 4 images of the classic PhotoDisc PDI demonstrating how large an effect improper hard proofing and use of M1 or M2 is as well as the much smaller effect of using M1 or M2 spectra profiles for normal printing.

A recent thread brought up the changes in ISO 3664:2009 (viewing) and ISO 13655:2009 (measuring equip).  Together, these change the recommended graphic arts viewing to be as close as practical to D50 including the D50 uV spectrum and associated instruments used to create profiles for hard proofing. Prior to this viewing booths had lower amounts of uV including some (LED types) with virtually no uV. This was combined with the lack of uV specification for the M0 (non uV cut) mode. There has been a long term trend towards using OBs in print runs and non-OB media for proofing. If, and it's a big "if," viewing booths had roughly the same amount of uV light as the spectrophotometer used to create profiles then there was no problem. Hard proofs matched. People were happy. But, being non-standardized, new tubes came out they were closer to actual D50 spectrally and then, suddenly, hard proofs didn't match press runs anymore. Not good. In the late 2000's instruments were introduced and ISO 13655:2009 established standards for implementing spectros that produced a reasonable estimation of the fluorescence from daylight uV (D50). Profiles generated with ISO 13655:2009 will now produce good hard proofs with non-OB proof media against high OB content press paper.

Keep in mind that hard proofing involves Absolute colorimetry because the hard proof is printed in such a way as to reproduce the target's white point. This has to be done when the proof media does not include OBs but the target media does or otherwise has a different white point value. Proof media is chosen that has a larger gamut, and dynamic range than the target media so that no gamut clipping occurs in the creation of the hard proof. In the following discussion I will assume the proof media is OB free and the target media has a high OB content.

One failure of hard proofing is having a much redder proof than the actual press print. This happens when the profile used for the press is made with M0 (or M2)  and understates the impact of fluorescence because of the lower uV content. The first image illustrates this.

One can also have a much bluer proof than target.  This happens when the proof is made from a target profile that used M0 or M1 but the proof and target are viewed in one of the older ISO 3664 booths that had little or no uV. This was especially characteristic of LED viewing booths which can be quite good from a metameric pov when only considering visible light. Here, the proof appears blueish because the proof has shifted colors to create the colors that would be viewed in a uV containing D50 illuminant. The viewing booth, not containing significant uV, doesn't shift the target's colors resulting in the proof looking bluish compared to the target. The second attached image illustrates this.


Now for the common issue where we use a profile made with M0 or M2. Ever selected M2 when you meant to use M0 or M1 and were surprised that you didn't see any difference? I have. There's a good reason the differences are so subtle.

Photographs are rarely printed using Absolute colorimetry and this results in a far smaller change between prints made with M1 or M2 profiles. This is because Lab (100,0,0) always prints to the paper white (look ma, no ink!) except in Abs. Col. So the print's white points will look the same whether you use M0, M1, or M2.

So what does happen is subtle shifts due to partial blocking of the paper's OBs as ink is added. If the area blocked by non-fluorescing pigments is large, a print made with M1 profiles will shift the b* negatively towards blue. This is typical of pigmented inkjets because the pigment is non-fluorescing. It shows up in my Canon 9500 II when there is a high light gray coverage as about a negative 2 shift in b*.

The third and fourth images are from M2 and M1 profiles as viewed D50 w/o uV thus the M2 profile image is the colorimetrically accurate one. The image from the M1 profile shows a slightly more negative b* than the M2 profile image as a tiny amount of blue is added to make up for the blocked uV. In the very light gray backgrounds b* is about a value of 1 more negative in the M1. I can see it if I strain hard enough and flip from one tab to another in Photoshop. Can I see it printed side by side? No.


http://everydaycolormanagement.blogspot.com/2011/12/tail-of-two-bulbs.html

[Tim Lookingbill]

Interesting write up, Doug.

Even though I don't build profiles I do see a correlation in the subtle color changes demonstrated even with my Epson "All In One" using Printer Manages Color on Epson Ultra Premium Glossy which looks quite bright and a bit blue with a tad toward green that requires me to apply a yellow bias to the source image to get a match under even my daylight fluorescent by reducing the blue channel middle slider in Photoshop Levels. If I don't I get more of the bluish bias in your posted sample. It's nice to see that the technology is this advanced to control that level of color bias in a proofing environment and actually see it even in consumer grade printers.

I had to A/B compare your PDI sample 3 & 4 turning on/off view layers in Photoshop where 4 does show a slight orang-ish red bias. The differences are more pronounced in the African American baby's skin just underneath the chin and across the chest.

What I don't understand or see the significance in that linked ISO discussion blog are the reasons for printing a hard proof with Absolute Colorimetric since the client is going to view under various lighting. Or is this for commercial CMYK press work, but then again why wouldn't the entire proofing workflow all have the same lights with the same amounts of UV content.

And on that subject do these new UV content ISO 2009 standards bulbs produce the spectrum of UV that can be harmful? Solux 4700K seem to limit their UV content above 375nm.

[Tim Lookingbill]

Here's what I get with my Epson "All In One" without yellowing adjusting the blue gamma slider in Levels. Note the OBA-ish white background behind the subject's is not as blue (filtered by the daylight flotube slightly yellow bias) but the African American skin is red but from the blue bias which is a flavor tint of cobalt blue, not cyan blue.

There's two types of red errors I see in slight mismatches where printer manages color on bright white hued paper and even with Fuji Frontier drylabs. It's either orange-ish red or magenta red (cobalt blue) but the white's are usually quite neutral until you see the duo tone effect viewing a grayramp.

[Doug Gray]

Interesting write up, Doug.

Even though I don't build profiles I do see a correlation in the subtle color changes demonstrated even with my Epson "All In One" using Printer Manages Color on Epson Ultra Premium Glossy which looks quite bright and a bit blue with a tad toward green that requires me to apply a yellow bias to the source image to get a match under even my daylight fluorescent by reducing the blue channel middle slider in Photoshop Levels. If I don't I get more of the bluish bias in your posted sample. It's nice to see that the technology is this advanced to control that level of color bias in a proofing environment and actually see it even in consumer grade printers.

Epson "Ultra" Premium Glossy has a lot of OBs whereas the regular "Epson Premium Glossy" doesn't though both substrates are quite fluorescent. When you print with RI, the colors will be adjusted, typically using the Bradford process, by the profile to match the paper's white point. Colors will be added or subtracted as required, through the profile use, to produce the images' adapted color that paper requires. OB's muddy things up because you can't see it and different viewing booths may or may not have "proper" uV levels. The specification of uV content in M1, which roughly matches the D50 daylight spectrum, is an attempt to stabilize hard proofing consistency in cases where the hard proof media has different OB amounts. In essence, the visible colors are shifted on the hard proof to match the visible colors created by uV fluorescence on the press run.

I had to A/B compare your PDI sample 3 & 4 turning on/off view layers in Photoshop where 4 does show a slight orang-ish red bias. The differences are more pronounced in the African American baby's skin just underneath the chin and across the chest.

Yes, the attached images do show an increased orangey red shift in the warmer areas. The subtle color shifts caused by OBs is toward the blue on neutral grays, and this is most apparent in rendering B&W images. The b* shift is 0 at L=100, changing to about -1 at L=80 then plateauing at about -2 L=60 down to just above the max density at around L=8. Overall, on a largely neutral image, the color cast is that of a slight  bluish one. I should run one of Bruce Lindbloom's very Lab check images as it then is easy to see how the exact shifts change across the L*a*b* range.

What I don't understand or see the significance in that linked ISO discussion blog are the reasons for printing a hard proof with Absolute Colorimetric since the client is going to view under various lighting. Or is this for commercial CMYK press work, but then again why wouldn't the entire proofing workflow all have the same lights with the same amounts of UV content.

Absolute colorimetric is required to recreate the exact paper color. A press has very different media to start with and a hard proof is required to reproduce the colors of the press media. Hard proof media is typically w/o OBs but still has a gamut that surrounds that of the press. This allows it to, for instance, add a bit of blue and touch of gray at RI values of Lab(100,0,0) which will be printed at an actual (AI) (L=91, -10, 0) on the press media's white point. In this example RI Lab=(100,0,0) is the same as AI Lab=(91,-10,2).

This matching was harder when the viewing booth uV and the uV content of M0 profiles didn't often correlate. That was the case previously. It's important that the side by side viewing of the hard proof and press run sample match. The new standards for spectros requiring M1 and  viewing booths requiring daylight (D50) levels of uV go along way to making that happen.

And on that subject do these new UV content ISO 2009 standards bulbs produce the spectrum of UV that can be harmful? Solux 4700K seem to limit their UV content above 375nm.

Even if the D50 uV was an exact match it's not much of an issue since the lux level in viewing booths is between 500 and 2000. Open daylight is well over 10 times stronger. Further, the booth user gets much lower lux levels than that because the viewing booth light hitting the observer is reflected light.

[Tim Lookingbill]

So you can't use these new UV content bulbs as a plant grow light, but those types of light designed for that purpose that I've seen clearly exhibit a pinkish blue cast so these proofing bulbs clearly show a neutral white, not yellowish, greenish or reddish.

[Doug Gray]

Can't say about Gro lights

Here's Lindbloom's L=50 chart which has a* and b* spaced at intervals of 10.
The first chart is just the RI of the Epson Ultra Prem. Glossy w uV as viewed under full D50.
The second chart is the RI of the Epson Ultra Prem. Glossy w uV as viewed under D50 w/o uV.

The differences are relatively subtle but the blueish patches are slightly more negative in the b*s. The neutral center has a b* of about -2. In the warmer colors b* catches up. By the lab (50,30,30) patch b* has no shift and becomes positive in the more saturated red/orange patches.

You can see (or measure) the differences by doing an A/B switch in Photoshop but they subtle.