Current Process (last updated 04/16/2008)
(because I've gotten lots of email asking)
Is this article useful to you? If you think so,
please consider a voluntary donation.
Contribute via Paypal
The Story
I've gotten several emailed requests that I describe my current process, my current choice of materials, and so on. In the past I've resisted because I think there are lots of perfectly good choices. Since I've gone to digital printing, I've been getting more questions - what film do you use? What developer? How do you scan negatives? What software?
The list just goes on and on.
Ok, here it is.
Caveat - this is what I PERSONALLY do. It works fine for me. Are there better ways? I'm certain there are. In some cases, I don't use those 'better' ways simply because they're too expensive. (No, I don't want to buy a $20,000 Imacon scanner. No, I don't want to buy a Lightjet. I'd like to buy a $35,000 medium format digital single shot back, but can't justify it.)
Every photographer's process is some blend of a desire for the best result and a compromise with practicality. If your situation is different from mine (where 'situation' is a code word for patience, attention to detail, tolerance for pain, disposable net worth, etc) then naturally your optimal process will be different from mine. That doesn't make you wrong (or me wrong), it just means that we've optimized for different things.
Image Capture
For everything I do that doesn't involve a hand camera, I use a large format, 4x5 camera. You can read all about the current equipment lineup in this article (The Large Format Kit). In brief, I use a 4x5 Linhof Technikardan 45s, with a variety of large format lenses.
Image capture is on film. With rare exception, that film is Kodak 100Tmax (aka TMX). I've evaluated the main competitor to TMX, Fuji Acros. Acros is a nice film, but it costs more than TMX so I use TMX. Sometimes I use Readyloads, sometimes I use regular holders, and sometimes I use grafmatic film holders.
When I was making exposures targeted for conventional silver printing, my exposure calculations were intended to produce what I call a 'robust' negative - one with appreciable density in all areas, including the darkest shadows. This was because TMX has a fairly short toe - if you underexposed and detail fell down on the toe of the film curve, the penalty was that that detail was lost forever, which limits printing choices. On the other hand, there is little penalty with TMX if you 'overexpose', because the film curve is nearly straight well past where you'd ever normally expose it. Highlight detail is always preserved, even with a grossly overexposed negative. In short, if you're silver printing, it's easy to ruin a negative by underexposing, but nearly impossible to ruin it by overexposing. For images targeted at printing on gelatin silver paper, my motto was always "when in doubt, add more exposure."
My exposure procedure has changed a little bit to optimize things for scanning. In general, scanners introduce noise into the scanned image. The amount of noise is uniform across the image, but it's more apparent in the densest portion of the negative/transparency. Unlike silver printing, there is a large noise penalty for having an overexposed negative. As a result, I'm quite a bit more careful when metering, and I'm also conscious of the tradeoff between avoiding noise in highlights and losing detail in shadows.
The film is developed in a Jobo cpp-2 processor, using Jobo 3010 Expert drums, and using XTOL diluted 1+3 single shot. I did a comparison of several developers and dilutions (including Tmax-RS 1+9, XTOL straight, and XTOL 1+3, and a custom divided pyrocatechol developer given to me by Denny Wagner). In terms of noise, resolution, and sharpness, it was a close call between the pyrocatechol developer and XTOL 1+3, with XTOL being the loser. But XTOL is easy to use and buy, so that's what I use.
In general, because of the noise issue, I'm finding that lower contrast negatives scan better than high contrast negatives. So I'm now generally trying for a negative contrast slightly lower than what I aimed for with gelatin silver printing.
The rest of the process is normal for B&W film - acetic acid stop bath, Kodak rapid fixer, Heico Permawash, and a 10 minute running water wash with a final rinse in distilled water, and isopropyl alcohol to promote drying and avoid sludge.
Wet, processed negatives are hung to dry from wires strung across my darkroom.
Moving into the Digital Domain (aka Scanning)
When dry, I sort the negatives into pairs (I expose two sheets of film for each setup, as a backup precaution). Once the pairs are identified, the negatives get put into polypropylene Light Impressions interleaving folders and the pairs are put into paper envelopes and have the notes for that image written on the envelope. At that point, I assign a unique index number to each sheet of film.
I then scan one of each set of negatives at 400 ppi, and save the scan in my index - a directory structure on my computer that lets me browse every 4x5 image I've ever made. The low res scans aren't suitable for printing, but it doesn't matter. They only get viewed on the screen. The index scans get stored as .psd files. TIFFs or jpegs might be better, but I use .psd files. The index scan is my equivalent of a contact proof - fast, cheap, easy, and good enough to be of use but not really intended to do much except allow the image to be evaluated so I can decide whether I'm going to ignore it or print it.
When I decide to print an image, I make a full resolution on my scanner, a Microtek ArtixScan 1800f. Full resolution means 1800 ppi. It's essential when scanning to avoid clipping away either highlight data or shadow detail. This is done by carefully adjusting the controls of the scanning software to ensure no clipping and to ensure that the output is nearly full range. I scan the negative as a 16 bit color positive, drop the green and blue channels and invert the scan in Photoshop. I do the scanning in color and convert to monochrome in this way because I get better results (see article here).
It's important to scan the negative in 16 bit mode, and to use repeated sampling to reduce the noise in the scan. For real prints, I use 16x sampling, which makes a substantial reduction in the noise level which is definitely visible in the final print.
I don't worry about compensating for curve distortion in scanning, because I always adjust the tonality of the image in Photoshop.
I use Silverfast for scanning. I used to use the Microtek software but some careful tests (see article here) convinced me that the Microtek software does a lousy job and throws away resolution.
It's possible to use the scanning software standalone, but I generally use the Silverfast Photoshop plug-in and import the images directly into Photoshop that way.
Image manipulation
After scanning, I use Photoshop CS to adjust the image.
I'm often asked why I don't use other image editing software. After all, Photoshop is the worst piece of software ever written, and it's expensive to boot.
But I use Photoshop because it's sufficiently powerful for my needs, and it's the de facto standard for such software. This means that I can trade techniques, experience, etc. with other photographers. For example, other photographers can read my article on toning of B&W photos digitally, can download the curves I provide, and can use my work directly. That's darn useful for them, and I'd point out that when they come up with nice curve files, they could probably send them to me, and I could use them, too.
And the books written on Photoshop would more than fill my work room twice over. Never mind that most of them cover such worthless topics as how to make your model's eyes (or other physical attributes) 15% larger so they look sexy; if you have 1% of the books that offer something worthwhile, you're far better off with 10,000 books available than with 10.
So I use Photoshop. I don't like it, but I use it.
I run Photoshop CS on a reasonably powerful (but not outrageously cutting edge) Intel based PC (2.8 GHz Pentium with HT, 1 GB of ram, a very fast 10,000 rpm 70 GB hard disk, and an array of USB 2.0 external hard disks for long term storage of images. The monitor is a Samsung SyncMaster 213T, profiled for me by a friend with his Gretag-Macbeth Eye One.
The two external hard disks are identical, not only in that they're the same model, but the files on them are the same. The risk of a single disk failure is not zero. The risk of two simultaneous disk failures is not zero, either, but it's far, far lower. So I keep everything important on two different disks. To show the level of paranoia I have, the files are not only duplicated on two different disks (two external Seagate Firewire/USB 2.0 disks) but also on my file server, a Freebsd based server running Samba to share a 1 terabyte (1000 gigabyte) RAID filesystem that consists of five 250 gigabyte Seagate SATA disks (four disks for data, one to provide redunancy. One disk can fail and the filesystem can be recovered without any trouble). That's like a belt, suspenders, and anti-gravity pants.
The first thing I do is spot the image. I use the cloning tool some, but mostly I use the healing brush. For large areas that will not need to be sharp, I create a quick mask to select just that area (the sky, for instance) and then apply the 'dust and scratches' filter to quickly spot large areas of the image. In areas that are not mostly large expanses of smooth tone, there's no other way to do it besides going methodically over the image and healing every little blemish. It's no more fun than spotting prints, but at least you don't have to spot every print. You just spot the image once.
When it comes to image editing, I'm a heavy user of Photoshop Layers. Virtually every change I make to an image is done with a separate layer, because it makes it easy to go back and tweak that change. I go back and tweak changes a lot - it's the way I work. One really nice feature that I use constantly is the ability to turn a layer on and off. I get the image as large as I can on my screen, and then I turn the layer I'm currently working on on and off. Which is better, A (on) or B (off)? Refinements of this can be had, as well. Want to know if you've gone far enough with a change? Duplicate the layer that implements that change, turn on the original, and then turn the dup on and off. Which is better, more or less? The History palette is useful in this way as well, and I use it heavily.
The first layer that gets added is an overall Curves layer, where I adjust the overall tonality of the print. The curves feature of Photoshop is, as far as I'm concerned, the primary feature. I now do burning and dodging and 'bleaching', etc. by creating a curves layer with a mask to restrict the area that's affected and then tweaking the curves to get the effect I want. In this way, you can burn, dodge, and control local contrast on very small areas of the image. You can burn or dodge just the highlights, by tweaking the highlight end of the curve; you can alter just the shadows by adjusting the shadow end of the curve, or whatever you please. Just remember that this unlimited power must be used only for good, and never for evil.
All of those manipulations are performed with the image in monochrome mode, 16 bits. When I've got the image the way I want it, I save a version of the file. Because I often go back and do more, even after I thought I was done, I save versions of the file as I got. The file names indicate both the image index number and also which version it is - thus, I'd have a file name like "050115-7a wip1.psd", which tells me that this image was the 'a' negative of the pair, from the 7th pair I developed on January 15, 2005, and that it's the first version (work in progress, or wip) of my ongoing series of manipulations of that image.
Making a print
After all that, I'm finally ready to make a print.
My prints are made on an Epson Stylus Pro 9600, an inkjet printer that is capable of making prints 44" /111.76cm on the short dimension. The printer is loaded with genuine Epson Ultrachrome inks.
In Photoshop, my first step is to flatten all the layers. Then, I adjust the size to be 360 ppi at the desired print size, using the Photoshop image size dialog. If that means upsampling or downsampling (it is almost always one or the other) I do that using the bicubic interpolation provided by Photoshop. I don't use Genuine Fractals, I don't adjust sizes in increments of 10%, I don't do any of that. I just adjust the size, in one step.
Yes, I've read all the articles on increasing image size in increments, and I've tried it, and to be honest, I believe that to a large extent, what's happening is superstitious behavior. Likewise, I've tried Genuine Fractals, and I don't find it to be a real improvement over just resampling with Photoshop; in addition, the interface for Genuine Fractals is just horrible. To read the reviews of Genuine Fractals, you'd think it was akin to some ecstatic religious experience, and I just don't find that's the case.
Once the image is the size I want, I do the unsharp masking. Generally, I do this with the image display set to 100%, and the sharpening is done around 100% radius 1 pixel, threshold varies but is low. In some cases (many cases) there are regions where I don't want to do any unsharp masking (for example, sharpening in the sky will just emphasize grain, ick), so often I generate a mask that prevents the unsharp masking from affecting those areas. In some cases, I'll even invert that mask and do a little blurring (very slight) in those areas that aren't sharpened, to make the tones a little more smooth. Sometimes, unsharp masking will reveal shortcomings in the spotting, and if that's the case, I do a little spotting work at this stage.
When all this is done, and I'm satisfied with the result, I convert the image to RGB mode (still 16 bit), and apply a toning curve to add subtle color variation throughout the tonal scale. This is described in my article on toning.
Now, finally, I'm ready to actually make the print. Currently, I print on Epson Premium Luster paper. I don't use a third party RIP, I just use the Epson Windows printer driver that came with the printer. I turn color managment off in the printer driver, and let Photoshop do all the color management. I use Bill Atkinson's free ICM profiles for the Epson printers - not only are they free, they're the best I've found.
Where do I go from here?
Remember, this is just a snapshot of the process I use. Like any photographer, I adjust this constantly, for reasons that vary from passing whim to discovery of real shortcomings. As I learn more, the process changes.
One change I'd love to consider would be moving entirely to digital capture - no film. Medium format digital backs are coming along nicely, and I suspect that the current crop of 22 megapixel backs are competitive with 4x5, in the same way the current crop of 11 megapixel digital SLR's are competitive with medium format film. The down side is that the MF backs are really expensive.
But I've been using a little Canon digital camera (a Canon PowerShot A95) to make photos for a daily photo weblog (see www.quotidianview.blogspot.com) and I've got to say, the freedom from processing film is a delight. Brett Weston once remarked that "99% of photography is sheer, brutal drudgery", and I'm confident that film developing was what Weston had in mind when saying that. The Jobo is better than using hand tanks, but it's still boring, boring, boring.
Other less expensive changes are in the works. It would be nice to upgrade the computer I use for image editing to a dual monitor system. I'll probably do that in the next year.
Likewise, I'll probably buy a Wacom graphics tablet, and see how I like that.
And that RAID file server - that's in the works, too.
But I'll point out that regardless of whether I use film, or capture digitally, the process is essentially the same as when I was doing conventional projection printing on gelatin silver paper:
- image capture - record the light intensities of the scene in some way
- image adjustment and manipulation - adjust that record of light intensities in some way, to better reflect what I want.
- contrive to have that adjusted image become permanent on some paper-like object.
That part will always remain the same, I expect.
