Well, actually a new printer except that it has set unused since getting unpacked and having the inks plugged in by its owner in December. The printer's history function is very clear: 0 prints, 0 ml ink used printing and 0 inches of paper.

What kind of printer keeps track of all that?

An Epson Stylus Pro 7900.

Drove to San Antonio to pick it up yesterday. A fair piece of driving for one day ... especially for me.

So far I've yet to actually print anything myself.

Gotta make room for it first.

That means that old things have gotta ... go somewhere.

This isn't a small problem.

In the same room I've got my Topaz 2 scanner (occupying 55"x30" of floor space), a desk (60"x30"), shelves (114" x 18"), more shelves (39"x10"), still more shelves (33"x13") and ... well, you get the idea.

It's like a puzzle whose parts can go together in different ways but only one way forms a coherent picture.

But it's been a great week!

Well, I've been set straight on some facts and this edit reflects it.

Sorry for any bad information.

Okay ... now I have a great scanner and so, of course, I'm thinking about a comparable printer.

I've been looking at the Epsons for their pigment based inks — which offer superior durability.

A well established issue with Cyan, Magenta and Yellow: combine two and you arrive at a decent Red, Green or Blue ... but mixing more results in progressively muddy colors. Thus black is needed to assure decent dark areas of a print.

On the other side of the scale, towards highlights, printer manufacturers have sought to get cleaner blends by using less intense colors — like "light cyan" or "light magenta". Also, additional neutrals lighter than black allows for better neutrals and near grays.

Anyone familiar with a photoprinter will be familiar with this list of 'colors', though the idea of a "light black" may be a bit confusing to some (it's just gray by another name).

So what happens when you mix yellow and light cyan? Well, it seems that you get a greenish yellow hue. Likewise, you can get orange hues by mixing yellow and light magenta. In fact, this system is great for producing secondary: red, blue, green as well as greenish yellow, orange, purple, and bluish cyan. Without a "light yellow" option included it might seem as if yellowish cyan or ruddy magenta might be off the table; however, anyone familiar with the Lab color space knows that yellows are bright by nature and you apparently don't need a "light yellow" at all. Thus certain Epsons up to and including the R2200/4000 Pro etceteras.

Another option is to add more primary colors so you don't have to, for example, mix magenta and yellow any more to get Red ... because red is already there in its own right. Also, having red around means it's easier to get clean oranges or ruddy pinks. Likewise, you can get the same sort of benefit by adding in Green and Blue rather than the light inks (save grays, because muddy is still muddy).

Canon's better inkjets include greens.

Now, the human eye is not a neutral instrument. We see more and better greens than either red or blue. Likewise we see more hues of red than blue. So if we have a Blue ink (as was the case with the R1800 and R800) our printer may have a comparably large color gamut (compared to a CMYK printer) but some of it will be lost on human eyes. With the R1800 the addition of Red and Blue seems to have been motivated by the desire to produce a strong spectrum of bright cool to warm colors seen in landscape photography — flowers and birds and all that.

The grass and the trees were left where they were.

But these are green: what we see the best.

So along comes the R1900 with Red and Orange inks in addition to CMYK. Some of the vivid cool tones (like bluish magenta) have been replaced by better greens (mix cyan with orange and you get yellowish green). Not sure why they didn't use Red and Green inks for the R1900. It may be that a greater blend of ruddy magentas is available with orange and the range of greens was deemed sufficient.

Of course, a true 7 color printer CMYKRGB, would be sweet. Especially if your toss in a gray or two.

Did I mention an 8 or 9 color printer would be much more expensive?

It's the old difference between a great design and a great product. A great design is a no-compromises triumph (any supercar or modified Roland 2x6 printer as a 12 ink uber-printer) while a great product gets the buyer what they want without taking their last penny (Mustang or an R800 printer).

Now for some speculation for the R2880 owners out there: your "vivid magenta" ... just what exactly IS it?

Actually, it's just ordinary magenta. I had originally speculated that it might be a reddish magenta but I've been set right by Jon Cone whose company make dye free pigment inks for printers (ColorCone Inks).

The R2400: when Epson made the R2400 they installed "cyan" and "magenta" in it that were not true cyan and magenta. The cyan was blueish and the magenta was reddish. This expanded the color gamut — the range of printable colors — at the expense of neutral hues. Good grays were to be assured instead by the use of three monochrome inks.

It turns out that the expanded color gamut of the R2400 didn't please professional people printing proofs, folks who could make a huge stink and cause problems for Epson.

Apparently, at least according to Mr.Cone, the suspicion is that marketing people believe a big color gamut equals big sales ... but one can make too many compromises chasing after a great product as opposed to a great design. At least with folks who are reasonably picky because their jobs depend on it.

What the R2880 does, an addition to other advances to regulate ink use and prevent clogs, is return magenta to what it was before the R2400. Since they don't hawk "Vivid Cyan" I guess the cyan used remains color shifted. This is unlikely to be an issue because humans simply don't see blues a strongly. Also, I would guess that retaining the "cyan" allows for better reproduction of blue-greens.

As with the R2400, the R2880 retains three neutral tones. With true cyan it can generate better neutral tones than the R2400 can.



So here's my revised thoughts: UltraChrome K3 with Vivid Magenta could be used in an older printer like the Pro 4000 and a custom ICC profile might suffice (because I wouldn't be trading out dissimilar colors). What would this accomplish?

Don't really know.

The Pro 4000 used the same ink set as the older Epson UltraChrome printers like the R2200. Did Epson "shift" cyan and magenta with these printers to get a wider color gamut ... only not quite as much as they did with the R2400?

Why am I thinking about these things? Because I want an Epson Pro 4000. Though I'm less sure than yesterday.

Yes: it's old. But it was also good and build like a tank.

As for my previous speculations about using different ink colors I've learned this: there's an expensive program called ColorGPS that lets you exactly do this. But there's no getting around the "expensive" part for me. Especially when used with old equipment like a 4000.

As I said befor, all I need is ... more money than I've got!

Well, I continue to work with my new-to-me Topaz. Straight out of the proverbial box I can report than my legacy ICC profile for Kodachrome (of which I have plenty to scan) is noticeably better than the standard profile shipped with Silverfast (which can be downloaded and used in demo mode, which is the full program except for painting "SilverFast" across your scans), which wasn't bad to begin with for being generic. This is nice since it means I don't have to be in any hurry to plunk down change on a Kodachrome IT8 target (LaserSoft wants on the high side of $200 for an individually measured sample ... which will yield the best results in calibration).






Anyhoo, I'm working up a methodology for scanning negatives that will produce superior results. It is actually an elaboration of the method I will now present.

If you've read about this problem before, or encountered it yourself, it all stems from that orange color cast that color negatives have.






Why the orange?

Well, early color negative films were like slide films and had a clear or gray base. If you were to look at a histogram of one of these films you would see the reds, greens and blues individually stretched all the way from 0 to 255 or 65K (depending on if you were scanning 24-bit or 48-bit color). The problem with this was that films and photographic paper are 3 dimensional objects no matter how flat and thin they appear to be, so as the light hits a sample of color print paper it passes through different sensitivity layers to achieve the desired effect. But each layer absorbs light so that lower levels within an emulsion do not receive the same quality of light.

Net result: prints from early color negative films had good blues, tolerable greens and sucky reds.

So someone figured out that by offsetting the sensitivity curves of color negative films, say by comparably increasing the dye density for reds, they could bias the negative to produce a good looking print. So when you look at the histogram of a modern color negative by color channel you see that R, G, and B appear to be compressed — no one color extending across the whole histogram — and all are offset from the others.

This is what causes the orange color and enables prints with vibrant reds and greens.

If only it were that simple then scanning a color negative would be as easy as defining a good "white point" and "black point", removing the color mask AND THEN inverting the image into a beautiful positive (with this method you don't invert the image to a positive until the color cast is dealt with).

And if you really want to get by, for mundane pictures lacking vibrant colors this method can work.






Here's the run down so far before I go on:

First scan the negative as an uncorrected color slide (preferably a slide type for which you have a good ICC calibration for your scanner). It is best to scan this in 48-bit color; however, you can convert 24-bit color to 48-bit in Photoshop if saving the scan as 48-bit outright isn't possible (more on "why" later).

Import this scan into Photoshop and use the Levels adjustment to select a "white point" (which in a negative is a place that will produce black on a print) and a "black point" (ditto in reverse): the orange color cast disappears instantly!

Here's some advice: you can use the eyedropper on an unexposed area for the white point and a light burned area for the black point as a quick and dirty option. These represent a sort of an absolute Dmin and Dmax for a film. It is better to find in-picture points to sample if you can because these do not stretch the data out as much and maintain a bit more realistic contrast (for example: with on picture of Copan's brother — Wailer — I chose a whisker and a spot respectively).

Once that's done you can invert the image and get an approximation of decent color.

More advice: you can save these adjustment settings, so smooth out a negative scan (but don't save it as changed) by blurring it a bit and then use the eyedropper. This means you could, in theory, even shop around among your negatives of a particular type (like "Kodak Gold 100") for the best white point and black point. You could also record the values from different negatives adjusted individually and either average them or else go with what seems to work best.

In the end what you are left with is a Levels correction that removes the base color mask and allows reasonably accurate highlights and shadows.

The next step to produce a good color is a bit trickier since it requires identifying a mid-tone gray — like cement — to represent a neutral gray balance for mid tones.

This step is necessary because the color cast does not have a linear response curve between the white and black points. This is where the ability to shop around among different negatives (of the same film type) may be important: few people include a standard Kodak Gray card in a picture that also includes white-whites and black-blacks.

Essentially, by identifying a patch of gray and grabbing it with the eye dropper you are defining at least one point to represent the deviation of the color cast in the midtones.

Choose these values well (remember, you can shop around in a negative with your eye droppers) and you have a basic color correction mask to use when you scan your family negatives.






Finally, why is using 48-bit color helpful?

As you work with a negative to remove the color mask you will see that you are expanding the ranges of each color so that their white and black points coincide. If you are only working with 256 levels of any color you can cause posterization (identified by a comb-like appearance to a histogram rather than by a solid X-Y plot). Posterization causes gradients that aren't smooth. Even converting a 24-bit image to 48-bit before you work and later flattening it to 24-bits for the final file is better than just sticking with 24-bits throughout.

Well, it's 8:16 PM on February 23, 2008 and I've just had my 4000th page view.

Cool

As people may have noticed by the last journal entry (now a deviation), I'm one to engage in a bit of legal loquaciousness at times.

Recently I've been in a discussion about different aspects of the civil war and I've been very careful in what I'm writing AND YET it seems as if I might as well resort to the lowest possible rhetoric for all the good my care and consideration has done so far.

For example, people have accused me of excusing lawlessness on the part of Confederates simply because I point out President Lincoln's own constitutional transgressions. I'm still debating how to respond even while I keep fighting the temptation to simply ASK THEM where I can find some offices of the Confederate States of America (or the phone number of my Congressman in ... Richmond, wasn't it?) where I can even challenge or question any wrongs that Confederate leaders may have done! Those offices (and phone number) don't exist because there isn't any Confederacy — period.

When talking about lawlessness in governance DURING the Civil War (not after or before) there are principally two different things to remember: first the Confederates attempted to take theirs "on the road" and they eventually failed in that effort — so the long term effects on governance of their lawlessness as Confederates died with that nation; second, other people had played fast and loose with lawful governance within the Union and we still live with the aftermath of those efforts in various ways. These are two different concerns and it is not necessary to excuse the one just to talk about the other. But try telling that to folks mired in rhetoric?

I'm starting to feel like Mr. Dent, I think like I'll go have a little lie down now.