The Jolly Green Negative

Friends-- With some trepidation I revisit this subject of the best blocking color (or ink combination) for smooth tones on digital negatives. And before I describe my tests, here are a couple of background items:

First, I am convinced Epson has changed the formulation of the inksets it supplies for the 3800 and for the 4880, sometime in the last year or so. In the old K3 inkset (as I used on my 3800) the strongest UV absorbers were dark black (pK or mK), Y, and LK. But C, M, LC, and even LLK were also significant absorbers and were useful. In the current K3 inkset only dark black (either pK or mK), Y, and LK are significant absorbers with the rest being at least 10-fold less absorbant. I have recently purchased the 4880 which uses the K3 inkset with vivid magenta. That inkset is very similar in UV absorption to the current inkset for the 3800. So, Epson has significantly degraded the ability of its inks to absorb UV. But I find we can still make good negatives with what is left.

Second, everything I have to say is only applicable to palladium emulsions. I have not tested any other.

Third, I have tried these ideas on both the 3800 and the 4880.

Now to the smoothness tests. In Photoshop I made a grayscale gradient 10 inches long by 2 inches wide, from 50% gray to 0% gray (white). Using QTR I then printed negatives of this gradient using different ink combinations. Of the various combos tested only three were really informative.

First I wrote a profile to print the gradient in just two inks, Yellow (70% ink limit) and Cyan (20% ink limit). (I attach the ink chart of this profile below). I printed this negative on a pure palladium emulsion. The print had two problems. One, as the print gradient got darker (ie, the Yellow ink dots were spaced further and further apart) the tones got rather grainy. Two, and a bit of a surprise, there was objectionable banding, parallel to the direction of the print head. So this pure green gradient negative flunked on both counts.

Second, Colin Graham kindly sent me a profile that prints without using any dark black ink but uses all the light inks except LM. I made a negative of the gradient with Colin's profile and a palladium print. This made a much improved print. The medium dark tones were much less grainy and the printer banding was greatly reduced (but not eliminated -- it would still be bad in a medium gray sky). If you look at the ink chart for this profile, I would attribute the improvement to the fact that the light inks are filling in around the Y and C inks. So I do not see this as a "pure green" profile.

Third, I printed the gradient negative using my current palladium printing profile for the 4880. The ink limits in this profile are:
pK 35 going to boost 55, C 25, M35, Y35, LC 35, LM 0, LK 40, LLK 40. As I said earlier, I think pK and Y are doing most of the lifting for the dark inks and LK for the light inks. To my eye the print from this negative was just as grain-free in the high tones as Colin's profile and printer banding was not visible.

OK, maybe adjusting my new 4880 would also help the banding (I will have to try) but clearly using multiple different inks will also cure it. As to grain, I have yet to find any ink combo that is more grain free than the semi-balanced inks in my current working profile. Another comment, I got printer banding on my old 4000 when I wrote profiles using only K and LK inks. The banding disappeared when I started using all of the inks. This seems similar to what I now see when I use primarily the Y ink.

OK guys and gals -- have at it. I expect my experience is not what others are seeing. But it is what I see and I can report no other.

Cheers, Ron Reeder

Well, that's disappointing that epson would change the inks formulation so often. The ink carts I've installed recently did come in different packaging- the recycled brown paper boxes- so maybe that corresponds with a reformulation. It's a little frustrating to spend the time pursuing an idea that may well be rendered moot by the next set of carts I install... Oh well, I've been a little stuck anyway- the RH here has been at 100% for a few days now, so haven't been able to test with my current carbon tissues- much too sticky in this weather.

But thanks for sharing your results Ron, and for taking to the time to mess around with my profile. I'm often disappointed how picky digital negatives can be from user to user, but at the same time greatly value the peer review on this site that exposes what is likely a dead end for the majority of users so quickly.

I'm going to quote you over on the hybrid neg thread so it'll will be there as a caveat for anyone who is considering trying the approach.

Serious news about the inks, Ron. That will cause a lot of grief with people's profiles... I'll run some tests - just ordering some new carts, so will get a comparison soon. What are the expiry dates on the "new formula" carts you are using?
Best, Ben

Ben Altman said:

Serious news about the inks, Ron. That will cause a lot of grief with people's profiles... I'll run some tests - just ordering some new carts, so will get a comparison soon. What are the expiry dates on the "new formula" carts you are using?
Best, Ben

Click to expand...

I am using some new cartridges and have not seen any difference in results.

On the other hand, I had big problems in updating to CS5 and Snow Leopard from Tiger 10.4.11 and CS3. Pretty well documented that you need to use some kind of work-around or there will be major color differences in results.

Sandy King

Ron-san said:

First, I am convinced Epson has changed the formulation of the inksets it supplies for the 3800 and for the 4880, sometime in the last year or so.

Click to expand...

I find this statement a little suspicious based on the fact that if Epson did that, everyone's carefully made printer/paper profiles (including Epson's) would be shot all to hell.

I won't dispute what you say but at the same time I'm not convinced that it is accurate. Seems if that did occur there would be thousands of complaints posted on the internet regarding the observed change.

Don

Is it possible that they were reformulated to match gamut, but not UV absorption? Maybe the pigment suppliers changed, or who knows what.

Colin could be right. Maybe they've changed the formulation of solvent/binder ect.

OTOH, Sandy's observation of getting no difference in terms of UV sensitive process printing is significant. I also didn't notice any difference in UV blocking of inks, while using PDN. BUT, since the driver has changed (I'm on Windows 7 64bit and CS5), and I had to change the print settings (I'm outputting Adobe RGB profiled files, using PS color management and I set the printer's profile to Adobe RGB, hoping this is equivalent to "No Color Management" of older system software/driver versions...), I get pretty different curves. I liked my old curves better!!!

Regards,
Loris.

With QTR you can cause the printer to print out each of its inks in steps from 0 to 100% of what the machine is capable of doing. I have been doing this for several years and then contact printing the printout onto a palladium emulsion. Recently I re-did this test using the Epson 3800 with matte black as the dark black ink. The resulting contact print is attached, labeled Test 9-11-10. (I don't think the 9-11 date has any significance!).

I then went back and pulled out a similar test done sometime in 2009. As you can see the UV absorption of the mK and Y inks was nearly identical between the two tests. LK was down a bit, but all the other inks were much less absorbent.

To the best of my knowledge everything else was the same between the two tests -- palladium emulsion composition, UV light source and exposure (100Units on my Nuarc), development, etc.

I also printed out each of the inks on my new Epson 4880 (theoretically the same inkset as the 3800 but with Vivid Magenta). As far as UV absorption was concerned the 4880 inkset was the same as the 3800. So, same result, two different printers.

All I conclude from this is that Epson, like Kodak and all other manufacturers, feels free to change the formulation of their products without any notice. Probably the change makes no difference in the visible spectrum. And unless you have been doing the sort of contact printing tests I do you might not have noticed it in the UV. Old ink or new ink, the major UV absorbers among the dark inks are Black (either pK or mK) and Yellow. And those have not changed. And LK is the major light ink absorber and it has not changed very much. And with all the fuss about handling profiles in the new CS4 and CS5 environments small changes would probably have been swept under the rug.

So the only lesson I draw is to keep an eye on your materials and test them now and then.

Cheers, Ron Reeder

Well I'm a bit skeptical. The density on the old black test seems to rise much faster than the new. It could just be the scans. Not sure if that's enough difference to account for the very different results in cyan and magenta. Almost like there was more restrainer in the old test. I use a bit of cyan as a filler, so I'm hoping it won't effect me much either way.

I have not had to replace magenta or cyan yet, so I can't compare old to new.

Hi Ron, I see what Rob means - the old test does gradate faster in all inks, as seen on screen, anyway. Is it possible that in the new test you did not have the ink limit at 100% in the QTR calibration routine? Or do the black and yellow strips in real life match in measured density on your new and old prints?

The only check I could do was with LC, which I replaced recently on the 9800 - 2012 expiry instead of 2010 expiry. An old chart on the UV densitometer looks pretty similar to the new one - less UV absorbed by the old, if anything. But I can't be sure the new ink has made it all the way through the line to the head yet, even with I think 4 recent head cleanings; it takes a while to get it through.

Returning to your original topic, thanks for the tests and post. You are confirming my belief that using more inks means using more nozzles and brings variation between the dot densities of each ink closer to each other - and both those things make for smoother negs and prints.

As I see it, the sensitizer reacts to the amount of UV energy it gets. It can't tell if the blocking is done by a black dot or a yellow dot. So if those dots are about the same size and spacing and block about the same amount of UV energy, the result on the print should be about the same. So a profile that uses a mixture of inks will be smooth if at each density the inks used each block about the same amount. I'm guessing, but this is just a guess, that "about the same" is fairly tolerant - a stop of difference or maybe more should not be objectionable, but three stops would probably be grainy. Also as you say, a neg of more thin dots has fewer gaps than one with fewer heavy dots.

Best, Ben

Ben Altman said:

Hi Ron, I see what Rob means - the old test does gradate faster in all inks, as seen on screen, anyway. Is it possible that in the new test you did not have the ink limit at 100% in the QTR calibration routine? Or do the black and yellow strips in real life match in measured density on your new and old prints?

Click to expand...

Uh-Oh. Boy is my face red. I took the old QTR ink page print out and measured the UV density of a bunch of steps of the mK, Y, and C inks. Then I did the same for the new QTR ink page print out (both printed with the 3800). Old inks and new inks measured very similar densities (especially so for the Cyan ink!).

So, I was totally wrong about there being a change in the UV absorption of Epson inks. Exactly how I screwed up the contact prints and/or scans I have not had the energy to figure out. But they clearly led me to the wrong conclusion. Deep apologies all round.

Howsomever, I still think the results of the smoothness tests are valid. Or at least until someone can show me where I screwed up there as well.

Moral to myself. If you have more than one way of checking a result, use them both before pontificating on the Hybrid site.

Cheers, Ron Reeder

Whew....

Yes, that is a relief.

Sorry, I was a little distracted and didn't notice what you said about the banding the first read- was this in my profile that you tried or just the 1-part yellow curve? I've not had any banding with my profile either in the printer output or the finished prints...yikes I really hope it doesn't show up.

A small point about your smoothness test- I agree for most processes that the full compliment of inks does work very well. But what led me to experiment with hybrid methods was high key carbon prints on metal, which is a very unforgiving surface for printer artifacts, on top of an already unforgiving process for highlights. I don't mean to assume that the hybrid negative thread prompted your test, but just wanted to point that out for clarity since you included my carbon profile in your palladium test.

As a friend who builds boats for a living once told me: "Everything we know we learned from screwing up."

So if you figure out what went wrong and it's interesting, let us know...