On digital negs and histograms
2 febrero, 2011 por
On digital negs and histograms
Jon Cone

UPDATE! We have made a new system called PiezoDN. Check it out at: https://piezography.com/piezodn/about-piezodn/


<Older Content>

Since posting the Digital Negatives blog, I have been asked by a number of people to explain why compressing a 265 gray value image into a smaller range of gray values will produce a high quality negative and subsequently a high quality print. Most tell me that everything that they have read has been to never throw away gray values. I would agree with them when imaging, or printing Piezography prints, or when making conventional inkjet prints.

The gray value requirements of a silver negative are actually quite shallow in comparison to a computer monitor display. So some background to this might be useful…

Let’s look at a poor quality printing system that has a limitation of being able to only separate and print 16 gray values. This could be a laser printer (perhaps). If we feed it an image file that contained 256 gray values – some type of compression would necessarily occur depending upon the software driver. So a smooth grayscale like this gradient would not print smoothly…

A smooth gradient

It would print something like this when viewed from far away..

16 gray values

And on closer inspection we might find that the reason it can not print more than 16 gray values is because it does not have enough ink gradations and is using some form of dithering or halftone effect at a low dot resolution. We might see something like this upon close inspection for a system that uses only black ink to produce a few gray values:

Halftone patterns in a black ink only laser print…

Even if this were a black ink only system, it could be used to make a digital negative although the dots would need to be incredibly small to look more continuous tone to our eyes. However, it would not have enough fidelity for most darkroom photographers. And the gamma or contrast of this process might not match that of our display system. Of course a simple curve might correct that. But, the 256 gray values are not adding to the fidelity of this system.

However, systems like this have been used on LVT film recorders as far back as the early 1980s to make silver prints. I used a similar system to make photogravure etchings from digital files in 1985. There is unfortunately a limit to how well black dots can fool our brains into thinking it sees smooth graytone.

Heidelberg presses at Steinhour

One solution is to use more shades of black ink. I think about the wonderful duotones and tritones which The Stinehour Press of Lunenburg, Vermont produced. These books were printed with multiple shades of black ink on Heidelberg presses using an incredibly fine halftone screen of 300lpi and which produced images much closer to continuous tone. They arranged the overlapping dots of ink in such a way as to avoid moire patterns and it was very difficult to not see continuous tone. But, as good as it was, it was still not darkroom quality.

Inkjet printers offer an opportunity for photographers to make their own negatives. Much of this pioneering work was done by Rob Steinberg of Palladio Company back in the 1990s. Rob used an IRIS printer and CMYK inks to make curiously green colored film negatives that closely matched the sensitivity needs of platinum paper. Epson printers and now HP printers are being used to make “color” negatives that work to absorb UV light for silver printing. Silver print is also sensitive to blue light and this is a different approach from what we take at this time.

Silver printing can use a negative that simply blocks light by virtue of being ever increasingly opaque. Silver grains act to block light. Silver negative film is a simple method that has been in existence for generations. Some photographers have learned to extend the dynamic range by staining their negatives with pryo. Some of the users of Piezography inks for negatives are opting for the Warm Neutral inks with their slight greenish/brownish stain which they believe will absorb some of the UV. However, we are also getting reports that the Selenium ink set absorbs UV.

Silver print itself, can not resolve the differences between 256 gray values. To be fair, neither can the human eye. The average human can differentiate far below 150 gray values at one time (in one lighting condition). The negative needs to correspond to the sensitivity of the silver print which reacts to densities between film base+fog and fully exposed film. This range usually does not exceed 1.85 on a densitometer. It can be made to exceed it of course, but the silver paper can not resolve details in such dense highlights.

Piezography can differentiate between 256 gray values because it prints thousands and thousands of gray values in order to do so. It can differentiate many more gray values than the 3 black system of Epson or HP because it adds two additional lighter shades of ink, and two additional intermediary shades between black and dark gray. (Piezography shades 4 & 5 correspond with Epson LK & LLK).

We design curves for Roy Harrington’s QTR driver so that we can control the dithering by making calls for ink only where the driver lays down so much ink that the dots actually touch each other, or do not separate enough to make visible dithering. We do not make darker inks appear lighter through dithering in the way the OEM must. Rather we use lighter shades of ink and we use a tremendous amount of back slope of lighter shades to fill in any gaps that might have paper white. This back slope of lighter dilution over darker dilution creates a great deal of intermediary values. We often overlap 3 or 4 dilutions at a time anywhere within the tone range. Either way, whether with extremely light dilutions or overlaps, Piezography is continuous tone from 0-255.

The Epson ABW dithering requires white space so that the dark grays (LK, and LLK) can be perceived as a half-tone by the brain…So their Epson ABW system produces dithering that includes a lot of paper white to appear “lighter” than the inks can produce. Of course they use some cyan, magenta and yellow in order to alter the color of their warmish light black inks.

this img src url From Digital Outback Photo’s Printing Insights #45 of the Epson ABW dither from an Epson 3800.

Side by side, Piezography K7 and Epson ABW looks like this:

Piezography K7 side-by-side with Epson ABW

Piezography is to inkjet printing, what 16 bit is to Photoshop. In Photoshop, a photographer working in 8 bit with 256 gray values can directly address each of those gray values with a curve or a selection tool. They could by example, eliminate all pixels with a value of 242. When working in 16 bit mode with more than 65,000 gray values, the photographer can still only address 256 values. The underlying engine uses the entire range of up to 65,000 gray values to smooth out gradients and redisplay the image to the photographers monitor or display screen which also is limited to 256 gray values. When the photographer prints to an inkjet printer from 16 bit mode, the output is limited by the device. Currently, no black & white output system can exceed that produced by seven shades of ink and the Piezography K7 curves for QTR.

We know from practical experience that 256 gray levels is all that is needed to display continous tone on a monitor (even from a 16 bit grayscale). But, those who use monitor calibration systems like Spyder or ColorMunki actually calibrate their video board and this usually and necessarily reduces the 256 gray output of at least two of the three channels of a video board. Photoshop is then necessarily displaying less than 256 gray values. But the point is not to see how few we can get away with, or how many we need to “throw away”. The point is to redistribute tone to a narrow range that reproduces a greater range when silver printed – and to do this without discarding or “throwing away” gray values.

If you’re still with me…the following digital capture taken on a foggy morning of an olive grove does not have 256 gray values. So we know already that we can make photographs in narrow dynamic ranges when the image necessitates it. Most systems can print this type of range without too much trouble provided there are enough light ink shades. Piezography would image this with thousands and thousands of gray values – created by the overlapping of so many carefully linearized ink dilutions. This type of output is close to the needs of the range of tone required in a silver negative. It is however represented as a positive instead of a negative. It is here to illustrate what happens when you put a silver negative on a light table and look at the very narrow range of density from film base+fog to fully exposed leader film. It’s simply, a very narrow range of density – and exceeding it only makes things harder once it comes time to printing it.

A narrow dynamic range of gray values…

So finally to gray values and compression…

Full dynamic range image…

After inverting the image in preparation for making a Piezography digital film negative the dynamic range of tone is simply reveresed.

Inverted image….

If the entire dynamic range of gray values is printed to the K7 film curve, the density range will be too high. The black will be far too opaque. Even the three quarter tones will be too dense for silver paper to realize any subtle separations. Therefore the end-points must be adjusted. And then a curve created that produces a negative that corresponds to the non-linear properties of silver. Those properties are affected by the process each photographer maintains. So there is no catch all curve. But the first step is to make a reproduction of a negative. That reproduction does not look like a full dynamic range image that has been inverted on a display. The corresponding histogram in a process like this does not reveal gray values thrown away, but rather as compressed.

Compression of dynamic range…

We need to print a range of tone when making a silver negative that is directly related to the needs of the silver print to produce a wide dynamic range. We’re not actually throwing gray values away. We’re compressing the tone range from the excessive range to the correct range. The contrast of a negative is quite low from dMin to dMax – yet it produces a greater contrast in the print. Dropping the Photoshop image from its 4.0 display dynamic range to that which is appropriate to a silver neg will result in a full dynamic range silver print. Otherwise, dithering of darker inks will be required and this tends to lower the prospects for con-tone.

Still I recognize that photographers will believe they are “throwing away” gray values if they can not understand the differences between compression and poor imaging habits. An example of throwing away gray values is what happens to a full dynamic range image that has been carelessly imaged.

The following example is what Photoshop experts lecture about and Photoshop books warn about…this is throwing away gray values….

Clumsily edited image results in loss of gray values….


A question begging to be asked….Has a photographer ever successfully printed and measured 256 distinct separations from a 256 step wedge that they had photographed, developed as film, and then silver printed?

A 256 step wedge is very difficult to obtain in order to photograph it, because it is actually very difficult to print one in any process that can separate measurably between 256 gray values. As luck would have it, a properly printed Piezography Glossy print of the 256 step chart we use for making profiles would serve this process (nicely I might add).

My concern around constructing digital negatives is that the silver film, while more sensitive than silver paper, can not actually differentiate between 256 gray values that we can see and measure with the devices we can afford. Certainly silver paper can’t. My thinking, and it may be totally off the mark, is that the digital negative needs to serve the limitations of the medium (in this case silver paper). On the other hand, one of the things I love most about printmaking is that there are so many paths to arriving at similar results.

In any event, the easy test is to photograph 256 measurably separated grays and attempt to develop a neg and print that can also separate them. I bet a dollar to a doughnut that it can’t be done. (If you try and you can’t, I prefer a maple cream doughnut – although a Krispy Kreme would be appreciated.) (If you can, send me the print and your density measurements and I will send you a dollar!)

Piezography when developed back in 1999/2000 could already exceed what the photographer could. At that time it used 4 inks and the Epson 3000 printer. In 2002, when I developed the second generation of Piezography we upped it’s linearization system. In 2005, we bumped its resolution by using more shades of ink. Those who have converted from the darkroom to Piezography have learned that it’s a greatly more sensitive system than silver. It remains to be seen if Piezography, by virtue of its tens of thousands of gray levels can improve the silver print. If so, how can it be more easily adapted?

I’m currently working on a new software that will linearize Photoshop to Piezography film. I have been in contact with someone else who is working on a linearization process that will linearize Piezography K7 QTR curves to film. Both of these processes are quite different and will present options to photographers who wish to make negatives and positives for their traditional processes.

On digital negs and histograms
Jon Cone
2 febrero, 2011