Decent negative scans the easy way
Table of Contents
OK, here's the setup: Having been a die-hard analog photo fan for many years, I've gradually gone digital. It all started with a Canon Ixus for snapshots and instant gratification for the kids. Discovering the drawbacks of reprocessing 8 bit/channel JPGs, I went along and got myself a Canon G9. And the fun returned to photographing! No more crappy analog-to-digital commercial lab work (“Naaah, that's not quite how it looked …”), no more dragging out the slide projector to be able to see pictures of more than half-decent size and colour balance. Back into the darkroom, only this time on the PC! Of course, also the G9 proved to have its drawbacks, so suddenly I was the happy owner of a Nikon D300. Being able to use my beloved classic manual focus Nikkor glass on the DSLR was a major factor in the decision process of whether to stick to Nikon or go Canon like my friends have done.
But what about all those analog pictures I've got from a couple of decades of analog photography? Naturally, they had to be digitised, so already by the time we had bought the Ixus, I “invested” (yeah, right!) in a film scanner. Enter the Nikon Coolscan V.
By the time the Nikon D300 DSLR was launched even I discovered that digital in many ways was as good as – or even superior to – analog. But how much better? I never was quite happy with my negative scans, they never looked as good as I wanted them to be. They just had to get better than that! Then I found out about Colorneg. And I read the hype. Already while reading about it, it seemed as if the program was written by a kinda geeky type of guy. Perfect! When I read about the calibration curves for different film types, I was hooked. Downloaded the demo version, tested for a couple of days and bought the serial number.
After two more days, I was mighty happy with the results. Gone sickly greenish or purple cast which I couldn't remove by Lightroom adjustments. So back to the question: How does really today's digital compare to digitised old-fashioned analog (or vice-versa)? I wanted to find out, so I shot a couple of pictures. Same scene, different equipment.
- Digital: Nikon D300, set to 12-bit NEF and sRGB colour space, using my old and trusted Nikkor 24mm f/2.8 AI.
- Analog: My beloved Nikon FE2 loaded with Fuji Reala, using my even older and slightly battered Nikkor 35mm f/2.8 AI.
For a review of the lenses I used, see e.g. Bjørn Rørslett's photo pages.
Yeah, yeah, I know. Different lenses. With the same lens on both cameras, I'd have to crop my negative to get the same picture angle, and I really wanted to compare on the basis of similar picture angle without cropping. Comparing DX format digital to FX format analog, I had to make some compromises. I chose to keep the picture angle constant. Bitch about it as much as you like, that was my choice.
I'd already bought Adobe Lightroom and discarded Photoshop (Elements or CS) for anything else than heavy image editing (which I didn't really enjoy anyway). The interface and workflow of Lightroom was just about perfect for an old-timer who'd spent a fair amount of hours in the darkroom and over the light table. Ergo, adjustments were to be done primarily in Lightroom.
Digital image processing
The raw file was imported into Lightroom, and a fairly rough optimisation in the develop module followed suit. I only used the basic controls.
Analog image processing by Colorneg
Using Nikon Scan 4.2, I scanned the image directly into Photoshop, using Adobe RGB (1998) as my working colour space. Scanner software settings were as recommended by the Colorneg people:
- Positive mode
- Gamma = 1.0
- No colour management
- Analog gain: R=0, G=1, B=2. Master gain was adjusted to stretch the histogram as much as possible without any of the channels saturating either into black or into white.
Since Photoshop had a nasty habit of crashing when I scanned more than one image at a time, I quit the scanner TWAIN module and saved the file as 16 bit/channel TIFF, as <filename>_neg.tif, after each scan.
I had already decided that I didn't want to spend too much time in Photoshop tweaking the image, so I applied the Colorneg filter without any other adjustments than specifying the film type. If the result turned out awful, I could always re-open the negative and try tweaking Colorneg for a better result. Already here, I was happier with Colorneg than with trying to tweak Nikon Scan to give me the best results. If the result turned out less than satisfactory with Nikon Scan, I would have to load and scan the negative again. And perhaps again. And perhaps again …
I then removed some noise and capture sharpened the image thus:
- Colour noise: Filter | Noise | Reduce Noise; Strength: 7, Preserve Details: 60%, Reduce Color Noise: 30%, Sharpen Details: 0%
- Convert to Lab, select L channel
- Luminosity noise was primarily present as single pixels: Filter | Noise | Dust & Scratches: Radius: 1 pixel, Threshold: 10 levels
- Sharpening of luminosity channel: Filter | Sharpen | Unsharp mask: Amount: 100%, Radius: 3.0 pixels, Threshold: 5 levels
- Sharpening naturally increased noise levels, so a mild noise reduction was applied: Filter | Noise | Dust & Scratches: Radius: 1 pixel, Threshold: 20 levels
- Convert to RGB
- File | Save As: <filename>.tif
- Exit Photoshop, import image into Lightroom
Analog image processing by Nikon Scan
No experiment is valid without the control sample, so I did a second scan using Nikon Scan to convert and adjust my image. Also this image was scanned directly into Photoshop, using the following settings:
- Negative mode
- Gamma = 2.2
- No colour management
- Analog gain: R=0, G=0, B=0. Master gain was adjusted to stretch the histogram as much as possible without any of the channels saturating either into black or into white.
Noise removal and capture sharpening was done the same way as with the Colorneg-inverted scans.
So. What were the results? Let's have a look. The picture was taken in early July 2009, in Lofoten in northern Norway. One of this year's holiday pictures. I spent a little bit of time in Lightroom trying to get colour balance, exposure and contrast as similar as possible for the three images.
I'm showing three versions of the images: The whole image downsampled to web resolution, a 1:1 crop from the centre of the image and a 1:1 crop from the top left corner of the image. The crops from the scans were slightly downsampled to the same resolution as the digital image. Cropping and downsampling was performed in Lightroom, the images were exported from Lightroom as 72 dpi sRGB JPGs at 90 % quality and output sharpened for screen.
Colour balance and overall aesthetic qualities
The all-digital image from the D300 was pretty good according to my standards. As if it shouldn't have been “pretty good” … A bit hard in the gradations and slightly lacking “oomph” in the colours, but pretty OK. I'm certain I'd get a better result if I'd spent more time tweaking the image in Photoshop or Lightroom, but I try to minimise the time used for tweaking.
The image inverted and adjusted in Colorneg really look nice to me. Nice colours, soft gradations. I used the Colorneg calibration without adjustments and spent perhaps a minute or so tweaking the colour balance and tone in Lightroom.
The image inverted and adjusted in Nikon scan is really not good at all. Sickly green hue in the shadows. Trying to remove it in Lightroom, I end up with a magenta hue in the highlights. You can see the clouds have a slight magenta hue already. Yeech.
Crop from the centre of the image
For the pixel-peepers, here's a 1:1 crop close to the centre of the image. The negative scans do have more noise than the digital image, but still the gradations are really nice in the Colorneg-processed scan. The rusty surface of the iron plate in the foreground has a more natural “rusty” colour than in any of the other two images, and the greens are really nice and warm. Just as they should be on a sunny morning in early July.
Crop from the corner of the image
Another pixel-peeper treat. A 1:1 crop close to the top left corner of the image. The negative scans still have more noise than the digital image, but interestingly enough, colour fringing (chromatic aberration) is more pronounced in the digital image. Even though the 24mm is considered to be a better lens than the 35mm. One up for the FX format.
Pretty simple: I'm happy with Colorneg and won't go back to fighting the NikonScan software to get decent scans from my negatives. I'm also pretty certain that my old and beloved analog equipment is going to see more use in the future, especially on those occasions when I don't want to lug one and a half kilo of DSLR, but still want more than using the digital compact.
Copyright © 2010 Størker Moe. All rights reserved.