The Big Picture
HSL RNP-Array - Double Cone
A Guide To RNP-Array Digital Negatives For Alternative Process Photography.)
Refining the negative Colour (Very important)
Handmade Digital Negative Curves
The methods and techniques discussed in these pages deal with making large negatives for contact printing. Alternative photographic processes require ultra violet light (light generated by sunlight or light from UV fluorescent tubes) and require a negative to be created which is the same size as the finished photograph. Currently, inkjet printers can do a very good job making these kinds of negatives on a transparent material called Pictorico OHP. By following the techniques outlined on these pages you will end up with a negative which is perfectly tailored to the "exposure scale" of the emulsion-paper combination you are printing to. BTW, these techniques can also be applied to "traditional" Black & White "Silver" photography. The advantages are that you can create a perfect rendering of your image on an application like Photoshop (levels, dust, dodging, burning, contrast etc.) and output a negative image which is contact printed to your photographic process and paper of choice. The big consideration for some here is the archival properties of printing to fiber paper and toning the print for permanence. The only limitation is the size of the printer and negative material.
There are two basic problems which confront those who wish to make their own digital negatives. The first problem is matching the density of the negative to the exposure scale of the emulsion-paper combination. The second problem is building and then applying a curve to distribute the available tones evenly when printing the negative. The first part of this site is devoted to matching density to exposure scale (RNP Arrays). In the "Handmade" section you will find information on creating curves by hand and by using Kevin Bjorke's ChartThrob script. OK, Onward!
I needed a cheap, reliable way to calibrate my negative's density range to my paper and emulsion's exposure scale properties for use in my alternative photography. The system had to use my choice of printer and inks. I had gone down the road trying other methods of finding colours for my negatives but the day I switched to third-party non-Epson inks my current system came up short -- or so it seemed. Let it be said that Epson makes really good inks. For the most part they block ultra violet light (UV) much better than aftermarket inks but they are also four or five times the cost. At $80-$100 a fill it gets expensive quite quickly to anyone on a limited budget such as starving artist wannabe like myself. Apparently, some of the "newer" Epson inks (the R1400?) don't block ink like the old Epson inks, this could be a potentially big problem for some people who are trying to make digital negatives using their current methods. For the record I use both genuine Epson inks when making prints and sometimes negatives. But I also use third-party inks for experimenting and when colour balance isn't as critical.
There were two breakthrough moments I should acknowledge. The first was discovering a simple technique that many others had already known for a while, which uses a new layer in Photoshop set to "screen" mode. This layer is then filled with a colour. I think I first read this on Brad Hinkel's original website. In fact, this is how I quickly built the grids for the original RGB versions of the RNP-Arrays. No, I didn't colour and fill the 800 plus little squares in the arrays (although, if I had to, I would have) They are actually two colour gradients overlaid on different layers -- much less work. It is also the technique used to "colourise" the negative creating the required density. The second breakthrough came when Clay Harmon published his elegant "Colour Ratio" method at alternativephotography.com. In trying to understand why it worked and then improve on Clay's original work I inadvertently taught myself to start working with different models and colour spaces. Finally much credit also has to go to Dan Burkholder for being one of the first pioneers to use the curve function in PhotoShop to fit the density range of the digital negative to the process being printed.
fig. 1 The Original RGB-RNP Arrays...Red-Green; Green-Blue; Blue-Red
Right-Click here to download tiff version. n.b. tiffs will better retain the colour information.
|The original three arrays are each a 17x17 matrix showing fully saturated RGB colors. Each array starts at 0 and goes up to 255 using steps of 16. When printed on to a transparent medium the different colours act like different optical filters when visible or UV light passes through them. In general, it seems Greens and Reds tend to be more effective UV blockers than Blues, but sometimes a little blue is a good thing too. Note: the ICM printer profile you select at the time of printing can bear a great influence on the way the inks are laid down sometimes completely changing the patterns of density in the print-out post exposure. Like everything else in photography everyone needs to get a handle on the variables involved with colour management (read don't change your printer settings)||To see a three-dimensional RGB Model click here. You may need to load a VRML plug in for your browser first, go here for that.|
If the above step wedges are printed to a transparency material like Pictorico OHP, or similar material, and are contact printed on an emulsion coated paper (e.g. silver, platinum, palladium, iron or gum bichromate) and exposed for a standard exposure time, a pattern of density will emerge as illustrated in figure 2. Simply stated the RNP-Arrays will match a colour to the correct density for your photographic process. Different processes require different density ranges to utilise the full potential of tones in the process. A negative made for platinum or palladium won't work very nicely if you try to use it for Gum Bichromate...and the opposite is also true. It's like fitting square pegs in round holes. Using a negative that is too "thin" for a process means you'll never see a decent highlight. Using a negative that has too much density means your shadows and highlight will always be muddy and blown-out.
The wedge below is a down and dirty get-the-job-done kind of approach. While not providing the big picture it will tell you quickly and economically whether or not your printer and inks are up for the job of printing digital negatives. The areas between the two black lines represents the limits of "pure" RGB colours.
Download the tiff version right-click and "Save As" here.
Different Emulsions and Their Typical Exposure Scales:
|Gum Bichromate||.8 to 1.2|
|Ware's New Cyanotype||1.8 to 2.0|
|Platinum - Palladium||1.5 to 1.9 (as high a 2.9)|
|Albumen and Salt||2.4|
fig. 2 Printout resulting from contact printing
|Here's my taped together cyanotype model of the "black" corner of the RGB cube. The three sides RG-GB-BR show how the density increases as the colours move into the black zone.|
|The HSB (or HSV) colour space has a lot going for it. For
starters I think it's nicer to look at than the RGB cube. It's based on the colour wheel concept
proposed by Isaac Newton so it's got pretty good legs for a
colour system. Each colour is assigned a hue based on its position in the
wheel measured in divisions of 360 degrees.
Complimentary colours are always plus/minus 180 degrees on the wheel. It's the
choice of many artists who work with colour. A nice feature is that
colours of similar hue are adjacent to each other. This makes it
possible to see how UV light is blocked based on hue changes. Unlike the
RGB model, and this is important, "Brightness" can also be
controlled by one of the three variables. This fact can be a big help in
finding the adjacent white-grey squares within the model. Wheels,
however, take up a lot of room when printed to expensive Pictorico OHP material so I opted for flat
projection which fits easier on an 8.5x11 sheet. The example below is a grid of
72 units across and 21 steps of "Brightness" deep to give a grand total of 1512 possible
Hue (colour) and Brightness (value) choices.
Fig. 3 HSB-RNP Array. Download Tiff Version of HSB-Array Right-Click and "Save As" Here.
Print output from the HSB-Array. It shows that even third party inks can easily produce the required density for classic cyanotype (1.0-1.2 logD).
The black line between the red and white areas is where blue and white squares are adjacent to each other. This means that for this particular process the ideal colour to match the density is found on the white side of the border.
Animated Argyrotype shot under the same UV lights (a little fogged at that perhaps, but you get the idea)
Above is an animated gif showing the relationship of density to exposure. It's interesting that you can see how some colours do a much better job than others at blocking the UV light. The top animation was compiled by exposing three cyanotype strips to UV light and exposing them at 10, 20 and 30 minutes. The densities measured on a step wedge were 1.2, 1.5 and 1.8 logD respectively. Of real interest is the hue located around 90 degrees which seems to reveal the existence of a "super-blocker" colour against UV light.
Another interesting thing about the above animations is that "spokes" or shadows of lighter density show up on the 300, 180 and then 60 degree columns. It's no accident that these anomalies are exactly 120 degrees apart. Pure Magenta, Cyan and Yellow. Now it's obvious these are the places where the CcMmYK inks on my printer don't overlap, so there is mostly just the ink straight out of the cartridge in those spots. Actually at 300 and 180 degrees two cartridges are probably firing M and Light m, and C and Light c. You would also think maybe the places where two or more inks overlapped (pure red, blue and green) would produce more density i.e. in the 0/360, 120 and 240 degrees areas, only this doesn't necessarily happen.
In general terms it looks like Yellow, or perhaps Yellow Ink, is by far the most effective ink straight out of the box, followed by Cyan and then Magenta. There is a definite "dense" area around the 90 degree mark though. I suspect this is because the wavelength of that particular hue is complimentary to the wavelength of UV light produced by my exposure unit. So I think really that two influences are at work here, dot overlap (mostly yellow with some cyan, equaling green) and density changes caused by the inherent UV blocking properties of certain colours. Another interesting observation is that "Black" ink (K) really increases density in the 0 to 20 per in the brightness range, near the bottom of the graph. My belief is that this is where the Epson print driver needs to rely on adding actual black ink to compensate for CcMmY ink's inability to produce a convincing black on printed output. This would tend to validate the belief using "black only" ink is a more difficult way to make a digital negative because the contrast changes so rapidly in this area.
The above HSB Array was shot onto silver MC Agfa paper under the same UV lights (for an extremely short interval) that I used for the Cyanotype and Argyrotype tests. Same pattern of density with the most blocking occurring at 60 degrees, then 180 and then 300. Now look at the same test of silver paper shot under my Beseler 45MX enlarger using a 150 W Tungsten light (no VC filter).
A new pattern of density. This would seem mean that the chemistry of the emulsion matters very little and the colour (or wavelength) of the light, in this case fluorescent UV or Tungsten, is the major factor governing what colour on the array is the most effective blocker. Different processes will be more or less sensitive to the light striking them (differing ISO speeds) but where light is blocked and the density produced seems determined by the light-to-filter relationship. See tests on MC contrast paper.
Another interesting thing when you use the HSB colour space to model the density paths, is realising there are more "usable" colours "inside" the HSB cone. The HSB-Array density strip shows just those colours on the "surface" or skin of the cone. If we were to start changing the "saturation" value the effect would be to start moving in concentric circles within the model toward the grey center. This gives HSB a big advantage over the RGB model because there's no simple way to do the same. Imagine a flashlight which can be focused from a wide beam to a spot focus, fully saturated colours are what you get when it's set to wide, but as you narrow the beam you get less saturation until the colours show only shades of grey. There would be at least two effects in doing this. The first would be helpful if you wanted to use colours which were pure grey in colour or that combined the printer's CMY inks in equal proportions. The second reason would be finding a colour that blocked light which wasn't fully saturated. Most likely this would be for a process which didn't require a lot of density in the negative -- such as gum. And finally one last possible reason for doing this would be finding a "smoother" ink combination which printed with less grain.
HSB Colour space shown as a disc (left) and as a cone (centre). HSL Colour model as double cone (far right).
RGB colour model shown as a fully saturated "Ell" panel (left) and as a cube with a cut away.
|The HSL-Array is a slightly
different way of looking at colour than the HSB/V models -- and I think
it is the most thorough of the Arrays so far. The
"L" stands for Luminance. In the middle-outside of the
double-cone this value is set to 50%. When the luminance increases the
hues in the model move
toward the white end of the double cone (top), when the luminance
decreases below 50% the model moves toward the bottom or the black
region. Like the HSB
model the most saturated colours are on the outside skin of the cone.
The reason the HSL double-cone could be useful is that some colours
available for density matching start in the "white" area above
the 50% line. Some Epson inks may be strong enough that the colours do
not have to start fully saturated. In the HSB model these colours would
be on the flat top of the model slightly in toward the centre. The
advantage in using a variant of the HSL model is the lighter colours can
be projected on a flat wedge to show the true starting point for some
processes which don't require a lot of density (i.e. classic cyanotype
The only drawback to this model is that PhotoShop doesn't have a palette option to work in HSL natively. Luckily, the HSB colour space is so close the Luminance and Brightness values can be easily interchanged. The red writing along the edge shows the HSL Luminance values with the equivalent HSB values in black ink. The best way to use this Array is to use the eye-dropper to sample the square that you choose as your negative colour.
If the above 3-D cones were skinned they would look like the HSL Array. Click here to download the HSL-Array in tiff form, here is the compressed tiff.
Update: May 2007
Now that I've had a chance to use the HSL Array I think it has by far the best potential of the three arrays. The first obvious benefit is that it shows the full range of densities for different colours giving the user a better understanding of where the their printer operates the most and least efficiently. Another benefit is that as you move vertically upward the HSL-Array mimics what your negative does as it goes from the most dense area to the least dense area. A judicious user can pick out "paths" which lighten in a predictable way similar to a step wedge. What I noticed on several test print outs were pools or "grey" areas which didn't desaturate very evenly causing the curve to flatten or curve upward. This would make using (or the dream of using) a single curve for all densities very difficult if not impossible.
An interesting discovery, or perhaps realisation, occurred while I was calibrating two colours to use in a comparison. I realised that the HSL Array is still an "orb" model even though I've projected it as a flat Mercator type map. At the outside centre there exists an equator similar to the equator on a globe. Any colour below this line must therefore pass into the globe to reach white (zero density) which is located in the white area at the top or what I like to call "the north pole".
It's unclear to me (at this point) if this is a major or minor (I think it's minor) downside to using colours below the equatorial line. The benefit of using colours above the line is that the HSL Array (as printed above at least) shows the colour fading to zero density, whereas there's no similar method for seeing that in the colour below. What I typically do after choosing any colour is fill a 101-step wedge to further pin point where I should be placing "white". At that point if there were to be extreme tone shifts (e.g. because of the driver's inking) I could do one of two things: ignore it and let the curve fix it or choose a different path and try again to get a smoother gradation.
Using colours below the line may be your only choice in some cases. So better to have some options rather than none.
Update January 2009: The path from saturated colour to white is a function of the type of colour fill method used. In the case of using the PhotoShop "Screen" fill the colour line followed really conformed to a pin being pushed into an apple. Interesting though I found another program called PhotoLine32 which can map a colour gradation more like wrapping a ribbon around a globe -- starting at the South Pole and proceeding to the North Pole. It makes using those dark lower colours much more usable because you can map them out with the HSL or B Arrays.
Above I have shown three colour models which can be used for matching a photographic process' density range to a particular colour produced by a inkjet printer. There really is no magic here. All I did was take the existing colour models and place them in the context of alternative photographic printing. I re-drew them with gradations which makes picking out a suitable colour easier.
To some extent the purpose in showing these models was an academic exercise in understanding the relationship between UV light, printer ink and various photographic processes. Some of these models may be overkill in finding an actual matching colour. Indeed, it is my hope that users will take and modify these models into shortened step wedges which can be quickly adapted to their working methods. Why print out the entire HSB or HSL Array if all that's really required is the 100, 75, 50 and 25 rows defined by brightness? It just means you have more time (and OHP) to spend making prints which is the end goal. Have fun with it. Drop me a line if you've found any of this useful.
-- Michael Koch-Schulte
Ready to make some Digital Negatives?.... Then click on, A Guide To RNP-Array Digital Negatives For Alternative Process Photography.
Questions or suggestions? Send to:
*What's RNP stand for? It's a recursive acronym.