One of the greatest things about being a film photographer these days is the ability to create images both on film, then scan that film into the computer and have both analog and a digital versions of the image: a hybrid work flow it’s sometimes called. There are plenty of reasons for doing this which others have articulated so I won’t veer off course here. Being mostly a color film photographer, a number of years ago I realized I was somewhat addicted to the brilliant color and punch of the chrome films. In a sense, chrome film was the low-hanging fruit of the color film shooter’s world – easy to scan because with the slide on the light box you began with a baseline of color to work with. Color negative film was a bit trickier in that regard – not being able to actually see the “right color” before beginning processing. But was it worth really diving in to figure it out? For a bunch of reasons I decided it was, and set about to figure it out.
I don’t pretend to be “the expert,” but I’ve been at it for a while and you learn what works and what doesn’t. Someone else asked recently if you print the image then lay the print on a flatbed scanner. This would make the scan a second generation away from the original; less desirable than a first generation scan from the negative. So the answer is no – you want to scan from the negative.
Assuming you already have a computer, you also need a scanner and scanning software. This description uses VueScan for the software, and the Nikon Super CoolScan 5000ED (also referred to as the LS-5000) dedicated film scanner.
The first step below is geared towards a Portra work flow, which I find myself shooting more and more. Again, lots of reasons for this – but the key is to over expose it by at least 2/3 of a stop or even a full stop. I’ve found when shooting and scanning all the Portra films, every one of them responds favorably when you give it more light than it thinks it needs. The question is, how much more? That depends on the scene, but some general guidelines are listed below. So with that in mind here’s how I get from a frame of negative film to a digital image in the computer:
Regarding the C41 (Color Negative) workflow, it goes something like this:
1) I almost always over expose color negative film by at least 1/3 a stop. So for Portra 160 I’ll shoot it at 100; for 400 I’ll shoot it at 320; for 800 I’ll shoot it at 640 or more recently 400. It’s not an exact science and your mileage may vary, but I do this because after a lot of experimentation with scanning have seen the most successful scans are on the light side, producing less noise in the shadow areas of the film. And if you’re worried about not getting true blacks, no fear: because you’re digitizing the file, bringing back the blacks is an easy levels adjustment.
2) I process straight. No push, no pull. So 160, 400 or 800 are processed at rated speed at the lab. I don’t develop the film myself, but send it to a lab.
3) When the lab process and sleeves, I have them cut the film into strips of 4 to archive in my PrintFile 35-7B sleeves which fit nicely in 3-ring binders. For those reading who don’t know this, the Nikon F6 generates EXIF data for every frame shot and captures it to a .txt file held in the camera’s memory. Unfortunately you need the pretty expensive Nikon MV-1 reader to extract it from the camera to a CF card, then import the tiny, data file into the computer. So I export the Exif data from the camera, load it into the computer, then import the .txt file into an Excel spread sheet. I have the spread sheet set up as a multi-page workbook and save the rolls off in chunks of 25. So one Excel doc will have 25 sheets in it. This streamlines the excel files as the number of rolls grows. Once formatted in Excel I print the sheet, three-hole punch and place it in the binder on top of the two pages of sleeves containing those negatives. Nice and organized, easily findable by walking to the binder on the book shelf, grabbing the right one and opening to the right page.
4) One by one I’ll review each frame of a roll on the light box with a loop to decide which frames I want to scan. My keeper rate varies wildly. But it’s usually higher than my digital photos.
5) Last year I switched to VueScan to run the Nikon LS5000 scanner. It takes a little trial and error at first, but I soon was able to establish consistent settings, allowing the maximum amount of data to be captured during scanning. For me, the magical settings switch between a couple different combinations, depending on the image. Most of the time I use what VueScan calls Auto Levels. In a properly exposed frame it does a good job setting the initial white and black point, and rendering semi accurate colors. Every once in a while, though, I’ll switch to the White Balance setting. I find VueScan’s color settings less than predictable, to be honest – and tend to baseline everything with the intent of working on color in either Lightroom or Photoshop. Because I’m scanning a pretty flat interpretation of the image and it’s 16-bit I know there’s lots of data there to work with.
6) I’m a believer in doing something once, and doing it right. To that end, I always scan at the maximum resolution knowing that once a master image is created, I’ll not need to scan it and rework it again-unless something unforeseen happens to the digital file. Scanning an image lower resolution, working on color, etc. thinking you can always scan it again later if you need it larger – is a waste of time. Storage is cheap, computers are fast. It’s best to do it once, get the master image exactly how you want it, then down sample versions from it. Down sampling also produces a tighter, cleaner, sharper file. To borrow an analogy, if you were to take a survey amongst 10 people, you would get a data point. If you were to increase the number of people who take the survey to 100, you’ll get a better representation of data. Increase the sample rate to 1,000 and even more so. You get the idea. So if you scan an image in at high resolution, then properly reduce it to a lower resolution, in theory it’s a tighter, more precise representation of the scene because you’re working with more data. So, I scan in 4,000ppi at 100% producing an image approximately 5,633 x 3,679 pixels. I scan right to the edge of the frame with the intent of getting every exposed pixel. The edges of the frame are uneven so I don’t always get every singe pixel, but it’s close. I used to scan into the black borders of an uneven frame, then rotate and crop and clone in Photoshop, but it was too time consuming and I’ve opted instead now to just let the rectangular selection determine the edges of the image. Sometimes I lose a tiny bit, but it doesn’t matter. The point is I don’t crop in scanning – but get the entire image at full resolution. Incidentally, this is one of the reasons I don’t have my chrome (slide) films mounted in slide mounts either. Especially when you’re shooting a camera with such a big, bright, 100% viewfinder like the F6, I like to get the entire frame exposed in the camera.
7) I scan at 4,000ppi because I’ve read in various sources and believe that’s about the maximum, uninterpolated resolution of a 35mm frame. You can scan higher on some scanners, but it may be interpolating or making up data as it scans just to get the higher resolution. I’d rather have pure, clean, raw information to work with than interpolated information. I usually use VueScan’s IR Cleaning filter, which does a great job eliminating the worst dust and scratches. I set it to Light, or occasionally Medium. I almost never use any of the other filter settings, but lately have been experimenting with the sharpening in VueScan vs. the High Pass method mentioned below. I never use Grain Reduction, finding that for me, it removes too much character in the image. One of the reasons I love film is the character the grain introduces – so trying to neutralize it is counter productive. If I need to reduce grain later I’ll do so in Photoshop with Noise Ninja.
8) I scan to 16-bit, TIF files, uncompressed, producing about 125Mb file. I’ll then open that file in Photoshop and begin work. Each image receives the following Adjustment Layers, usually (but not always) in this order:
a) Often the first thing I’ll do is duplicate the base image to make a High Pass layer. This finds and sharpens the edges within the image, affecting the tones less. The pixel radius setting is resolution dependent so for the full resolution image I’ll often play with between 2 and 3 pixel radius and see which works best, then set the Layer Mode to Overlay. This produces a non-destructive sharpening of the image that can be layer-masked to taste, with layer opacity adjusted independently.
Unfortunately however, it doubles the size of the file, and at 16-bit you’re now working with a 250mb file. So unless you have a pretty fast computer you may reduce the image to 8 bit at this point. If the High Pass setting is too high it’ll produce an awful white halo around objects in the final image so best to be conservative.
a) Levels: Next I make sure I have the black point and white point where I want them. To do this, I’ll put my finger on the Option Key (Mac) and drag the black slider in the levels histogram until I begin to see clipping. Sometimes I’ll push into the clipping a bit if I want a darker rendition of the image, other times I’ll just barely touch it. I’ll then do the same thing for the white point: finger on Option Key and drag the white slider in until I begin to see clipping over all, or in various channels. If need be I’ll also adjust the middle slider to work the gamma. You’re affecting contrast with this step.
b) Photo filter: sometimes a warming filter to bring up a naturally cooler image, sometimes a cooling filter to bring out the blues. The filter depends on the image. Sometimes LBA, sometimes Warming 85, Cooling 82, LBB, etc.
c) Either a Saturation or a Vibrance layer and work on the color. In most cases I don’t saturate above 15 – 20 points. Just enough to pop the color, but at least on my monitor, not into the garish zone. Sometimes I’ll push a little more.
d) Depending on the image I’ll put a brightness/contrast layer on to work the balance of the final image.
e) Depending on the image, it may require layer masks for any of the above layers or additional adjustment layers – effectively dodging and burning various portions of the frame. I’ll typically use one mask per adjustment layer and set the opacity of my pen tool 33% to 66% ( 1/3 stops) to build up a mask in successive strokes – not just use black. I always use a Wacom tablet and pen, finding a mouse far too cumbersome for such work.
8) I’ll use the Spot Healing brush and clone tool to remove the obvious scratches, film buggers or dust spots that VueScan’s IR cleaning filter missed. The goal for each image is to make the scan appear as close to the original scene as possible. To that end I don’t retouch images beyond simple dust and scratches, and color/levels settings.
9) I’ll save off the full-resolution, 16-bit PSD, then create a lower-resolution 16-bit TIF version (with layers intact). For electronic presentation I have a set 3:2 aspect size (1,800 x 1,200 pixels) and like a modest, white border around the image, with small, unobtrusive text containing the location and date of the image below, flush left and right with the edges of the image. I have created frame templates for this basic presentation: one for vertical, one for horizontal. The goal is to have the images appear essentially the same size and shape so it’s the content of the frame the viewer is seeing – and not distracted by the shape of the image. I very much like the 3:2 aspect and stick with it. When I downsample the image I use Photoshop’s Bicubic Sharper setting, seeing a difference in the results over the other methods. For a long time I refused to crop the images at all in this step electing instead to set the dimensions at 1,200 x 1,8XX (whatever the organic width or height was), but have since relaxed that in favor of creating a standard, repeatable and consistent size/shape. In a case where cropping off the 30 or so pixels at the edge of a frame will hurt the image, I’ll leave it.
10) Once the image is at it’s final size (resolution) I’ll sharpen. I used to use the LAB method: convert the RGB image to LAB color space, grab the Lightness Channel, sharpen, then reconvert to RGB. I don’t do this anymore with Photoshop CC because I’m really liking the additional features within the Smart Sharpen filter. With Smart Sharpen often I’ll work the Amount, Radius and Noise sliders to a suitable balance. If I’ve done everything else correctly the settings remain pretty low. If not, here’s where I’ll try to fix it – but abhor overlay sharpened images. Typically the Amount of between 50% and 100%, a radius of .5 pixels and a reduce noise setting anywhere from 10% up to 30% or 40%. Sometimes for Portra 800 I’ve gone up to 50% until the overall effect is smooth and pleasing – without producing a overly de-noised image (at least to me).
11) The PSD is saved to a directory on the HD created before scanning. After much trial and error I’ve settled on specific, unique code for each roll shot. It goes something like this: F6-r0265-POR160 is the directory’s name. F6 identifies the camera. r0265 corresponds with the roll number in the camera-generated EXIF data. This is important when marrying up the individual frame’s shooting data later. POR160 is a 6-character ID capable of identifying the emulsion of the roll. Each image that goes into that directory then begins with this code, then adds an fXX for frame number. So this image is F6-r0265-POR160-f25a.tif. This naming convention allows me to search the entire computer’s 8TB pf storage for a unique, specific image at any time and produce the PSD, the TIF or the JPEG. I can also locate any EXIF data file.
12) I use Lightroom to catalog all images, leaving them in place on the HD and referencing their position. I’ll let LR stack the images so there only appears 1 image in a stack of 3 version. This keeps the overall LR presentation down as much as possible. More lately I’ve been using Lightroom’s excellent processing tools to rework certain images after everything above has been done. I can see perhaps using LR exclusively in the future.
13) When I save off a JPEG for a specific use, sometimes I’ll re-size it. I’ll then throw it away after I’ve uploaded it for its purpose. I don’t like having a bunch of duplicate JPEG’s laying around.
That’s it. One image can take anywhere from 5 minutes to 15 or longer, depending on how much fiddling I need to do. The goal always though is to make it as good as it can be in the camera, understanding that the scan usually requires work to realize the images full potential. Some think it’s crazy to spend this time scanning film images in this day of digital everything. I guess I don’t agree with that – but respect the individual’s decision to use whatever means which speaks to them best when it comes to creating art. For me, it’s still film and will be until they stop making it.