Scanning color negative film is something that often throws people for a loop when it comes to getting “good” colors. I can relate – I still find it a frustrating endeavor sometimes. And it isn’t helped by the algorithms in scanners and scanning software that are intended to make this process easier. The actual result is that these automatic color adjustments make life pretty frustratingly difficult if you’re after consistency! Here’s an illustration (and a solution, but I already discussed that earlier).
Let’s have a look at a series of five consecutive Kodak Ektar frames shot on a bright and sunny day in European summer. They’re not exactly Ansel Adams captures and a fairly random film strip that was sitting on my desk.
This is the strip as scanned by my Epson 4990 in positive mode without any adjustments. It’s a reasonably straight representation of what the film looks like in real life. Perhaps a bit dark, but this is due to the fact that the narrow density range of the color negatives occupy only a tiny bit of the dynamic range of the scanner. I’ll come back to this strip later on.
First, let’s see what the scanner makes of these frames if I scan them as color negatives with minimal automatic adjustments. I scanned the frames individually, which would be representative for a typical end use application: you sit down, select a couple of frames you want to scan and have at it. Here’s what I get when I do that and assemble the individually scanned frames back into a single strip:
For clarity’s sake, here are the settings I used in Epson Scan:
I could fiddle with the settings a bit; for instance, I could disable “Continuous auto exposure”, or enable “Color Restoration”. Surely, this will make a difference. One thing that doesn’t change, though, is that the individual frames will come out balanced differently. Note the first three frames in the strip above: they were shot under the same light, within about 1km distance from each other in a timespan of about 10 minutes. Yet, the colors are pretty much all over the place (and not a particularly good place, either…)
The problem becomes apparent if I go back to the single positive strip scan I showed at the start of this page and then manually invert and balance the colors for all frames at the same time, as I described in an earlier blog. Here’s what I get when I do it that way:
We can debate aesthetics, personal preferences and whatnot. Maybe you’d prefer these to be a little brighter in the highlights, or a little warmer in terms of color balance. Fine. And I’ll readily admit that the exposures are kind of…ehm. (These were happily snapped with a lovely little Olympus OM10, which despite its many merits is not exactly stellar in the metering department.) The first three frames might have benefited from a UV filter on a day like this. Sure. And if I were to use any of these individual images, I would adjust the curves on an individual basis to make the most of it – to make the clouds pop a little more on the first frames, and to reel back in the highlights on image #4, etc.
There’s all that. But at least, the color balance between all five frames is consistent. The differences you see, are differences in the quality of the light and the subject matter. The differences we saw in the “let Epson handle colors for me” example were mostly differences in how the software interpreted the images – which wasn’t very successful, if you ask me.
Maybe we can throw another scanner into the battle – one that’s more geared towards handling film to begin with. Let’s be honest – the Epson 4990 is a nice machine in its versatility, but it’s still a flatbed that excels at scanning reflective media, not necessarily film. From the same era, I also have an old Minolta Scan Dual IV. Let’s put that one in color negative mode, so it inverts the images for me and gives me the colors it deems appropriate. Here’s how the assembled series of five frames looks like in this case:
What stands out immediately is that the Minolta has a totally different idea about what good colors are than the Epson. You may or may not like it (I don’t, really, to be honest), but it has this warm and fuzzy thing going on. Consistency seems a little better than what Epson gives is, although it’s still not great if you look at the colors in the skies on the first three frames. So here, too, the scanner decides on a per-frame basis what might be an OK result, and consistency goes out of the window.
What happens if I use this Minolta scanner to scan the color negatives in positive/slide mode? Here’s the end result after assembling the series together, inverting and manual color balancing all frames in one go as per my previous blog:
It’s actually not bad in terms of how the first three images now approximate each other much more color-wise. But I have some doubts about how warm frame #5 turns out despite the light not being so much warmer than in the first few frames. It seems to me that in positive mode, the Minolta still does some frame-by-frame auto-adjustment. This is the Achilles’ heel of a scanner like this, which can only scan one frame at a time, and that has limited options in software to fix color balance. This is where using something like VueScan might actually help some, at least if it’s set up to disengage any and all automatic color and contrast changes between scanning individual frames.
For fun, let’s make a collage of the four versions of these frames. This emphasizes the differences in interpretation, and also the difference in contrast between a ‘positive’ scan on the Minolta (with lots of auto-contrast adjustment, I suspect) and the Epson (which is a pretty consistent flat scan in my experience).
- First row: Epson 4990, scanned as negative, automatic color balance
- Second row: Minolta Scan Dual IV, scanned as negative, automatic color balance
- Third row: Epson 4990, scanned as positive, manual inversion and color balance
- Fourth row: Minolta Scan Dual IV, scanned as positive, manual inversion and color balance
One thing I did not (and will not) do, is to add a fifth row, which would be the same frames printed optically onto RA4 paper. I can assure you that the result will be yet different, even if I tried to color balance as closely as possible to any of the four frames in the example.
It is sometimes said that with color negative film, the colors are somehow ingrained in the film itself and determined by how the film renders them. It’s true that a film has of course certain characteristics in terms of how it will record different wavelengths of light, how linear it is, what kind of crossover it will give etc. However, when it comes to color balance, there’s a massive amount of interpretation going on.
If inversion and color balancing is left to the scanning software, especially software that tends to come with the scanner, there’s a lot of variation on a frame-by-frame basis. I think it’s inherently difficult (impossible) to put the final colors down to “the character of the film” or something along those lines. It’s the film, but also (and very dominantly) the character of the scanning software.
My recommendation remains that if you want good, consistent colors from color negative scans, consider the following:
- Determine what “good” is, to you. There’s no absolute benchmark that’s somehow imbued into the film. Color negative film has always been flexible, color-wise, like a modern digital camera RAW file.
- Consistency requires that per-frame automatic adjustments need to be disabled. With a flatbed scanner, this can be done by scanning all negatives in one single pass and then manually inverting and adjusting colors to taste, as described in one of my previous blogs. With other scanners, you may have to fiddle with the software that came with the scanner, or just push that aside and use something like VueScan that allows you to take more control over signal processing.
- If you want real-life accurate colors, consider photographing a color checker chart and keeping that actual chart at hand when color balancing the final scans. You’ll notice it’s very difficult to get an exact match between your (calibrated!) monitor and the real-world card under (broad spectrum, high quality!) white light. Don’t fret over it – if you want dead-accurate colors, be smart and shoot digital.
Of course, a good alternative to all this dreadful bit-pinching and byte-twisting is to take your color negatives into the darkroom and print them optically. If you print each frame with the same filter settings (and you did a reasonable job at exposing them properly), they’ll come out perfectly consistent, color-wise. You may also find that once you get the knack of dialing in appropriate yellow and magenta filtration, it’s a whole lot easier to get naturally looking, pleasing colors when printing optically than when doing scans. At least, that’s my experience.