When ‘Cinestill’ film came around, I just had to try it – although I have mostly used the original Kodak Vision3 stock instead of the remjet-less Cinestill variants. Frankly, I didn’t get along with it, and I’ve been quite vocal about it on the forums, too. But a couple of years after my initial bout of experimentation, I felt it was time for a re-assessment. Here’s a brief reflection on color balance of Kodak Vision3 250D film. After all, color balance is perhaps the most pressing matter when it comes to using Vision3 films for still photography.
Without regurgitating what is explained elsewhere on the web at length, let’s look at a few notable differences between Kodak Vision3 film vs. regular C41 film:
|C41 color negative||ECN2 color negative|
|E.g. Kodak Gold, Ektar, Portra films||E.g. Kodak Vision3 films|
|Color negative film with orange mask||Color negative film with orange mask|
|Intended for still photography in formats of 35mm up to 8×10″||Intended for movie capture on 16mm up to 70mm|
|Designed to be developed in C41 color chemistry; developing agent CD4||Designed to be developed in ECN-2 color chemistry; developing agent CD3 (like E6 slide film or RA4 paper)|
|Contrast and color balance originally designed to match RA4 color paper||Contrast and color balance designed to match movie-stock printing films|
|Regular anti-halation dyes||Carbon-based remjet-backing with anti-halation, anti-scratching and anti-static properties|
So let it be crystal clear that these families of products have totally different intended purposes. One of the surprisingly few things they have in common is that they are both color negative materials with an orange mask (read here why this mask is there to begin with).
Currently, Kodak is the only firm that manufactures ECN2 motion picture films, and their current product offering consists of a couple of flavors of their Vision3 product lineup: the daylight balanced 50D and 250D, and tungsten-balanced 200T and 500T. Back in the day, FUJIFILM used to make their own Eterna product line, but those times are long gone.
As indicated in the table, a notable difference between still color negative (C41) and motion picture color negative (ECN2) is the development chemistry, and in particular the developing agent. It’s the more modern CD4 for the C41 process and familiar films like Kodak Portra and FUJIFILM Superia, and the slightly older CD3 for the ECN2 process films like Kodak Vision3. Coincidentally, CD3 also sees use in the development of RA4 paper and E6 color transparency film.
The significance of the difference in developer is that the actual dyes that form the colors are chemically quite different between C41 and ECN2 films. The dye couplers in both film types are engineered to produce the desired color absorbance and stability when developed with the correct developer. Cross-processing is certainly possible – i.e. you can actually develop ECN2 films in C41 chemistry, and get a color negative image. It even looks fine at first glance. But it’s not exactly as Kodak intended it to be for its original purpose. Is that a bad thing? That’s up to you to decide.
And there’s the personal bit coming in, because what you make of all this as a still photographer, is indeed quite subjective. If you use ECN2 film for still photography, you’re essentially leaving the path laid out for you by the manufacturer and that means you’re pretty much on your own when it comes to the performance of the product. Some find that, and the challenges it comes with, a deal breaker. Others find it a welcome diversion, or at least find the penalties acceptable. I’m still on the fence, to be honest.
So what are these challenges in using an ECN2 film for still photography? In practice, it boils down to just two: (1) Dealing with the remjet layer, or accepting the notable red halation around bright highlights you get when using film without that layer (e.g. Cinestill films). And (2) dealing with the contrast, color balance and color crossover compared to C41 negatives. Let’s put #1 aside for a bit and have a closer look at #2.
For me, personally, the main objection against ECN2 films for still photography has been the challenge of optically printing the negatives onto RA4 paper. Since the contrast and color balance do not match the response of RA4 paper, there’s a fundamental problem here. However, since my initial foray in this territory, I have learned two things that make me re-evaluate all this.
Firstly, I’ve seen (scans of) prints that I find sufficiently credible and convincing that decent optical prints from ECN2 films should be possible. My own testing a few years ago wasn’t particularly successful, with especially substantial blue/yellow crossover as well as a more moderate green/magenta crossover as well. But maybe that was a fluke, I was doing something wrong, or…I don’t know what. Sometimes, giving it another go a few years down the line somehow resolves problems that existed before – much like my present carbon transfer work, that is lightyears better than what I was capable of before. People learn, and I’m fortunately no exception.
The second thing I learned since then is that RA4 paper basically doesn’t match any color film in existence anymore, since it’s optimized for digital exposure anyway. Given this, I thought at some point “what the heck, if nothing works, then it’s all the same in the end.” Instead of abandoning the whole thing, I figured that analog color printing can be more of a creative endeavor and that accurate color reproduction perhaps shouldn’t be the primary target – at least not for me. And that gives the freedom to experiment and perhaps hit upon effects that are entirely unrealistic and unintentional, but still worthwhile. Well, let’s see how that goes, because I have to admit I find it hard to NOT try and print straight!
Long story short, I got myself a nice big roll of Kodak Vision3 250D to play with. Having spooled a couple of test rolls for initial experiments, I could go out and give it a go. The first two experiments I can already report on here.
By ‘experiments’, don’t expect anything scientific. Not tightly controlled scenes, rigorous testing etc. Sure, that would have been useful, but also mind-numbingly boring.
Both rolls I exposed at ISO 250 – it’s what the film is rated for by Kodak, after all, so why not start there? Moreover, ECN2 film is engineered to produce a lower gamma than C41 films. I compensated for this in development, so that means I should at least have no speed loss in any case. On the other hand, I’m skeptical at the speed boost that firms like Cinestill claim by marketing e.g. 250D as a 400 ISO film.
The first roll I developed in ECN2 developer, but with C41 bleach and fixer. I DIY-ed a simple ECN2 developer and significantly overdeveloped the film for 4m45s at 41C. This was probably a bit too much, as contrast was pretty…er, firm. In darkroom printing, I had to use preflashing exposures to curb the contrast – see e.g. here for an example. I also did a further roll for a more sane 4 minutes of development, but have yet to assess these negatives.
The second roll I decided to cross process in C41 chemistry altogether, at C41 times and temperatures. ECN2 films will give a significantly boosted contrast this way. In fact, I found the contrast to be comparable to the first roll. The scenes are different, so the comparison isn’t exactly straight, but given how the negatives print optically, the rolls were comparable in this respect. In other words, my impression is that C41 development of this film gives too much contrast for convenient printing.
The scans above suggest that decently looking results in the digital domain can be obtained from both films. I did not try very hard to color correct the scans above. I did what I would normally do with color negative film: tweak the individual R, G and B curves a bit in the Epson Scan dialog until the preview looked close. Then I did some more minor adjustments with curves in GIMP, this time only manipulating the G and B curves – conceptually similar to what I’d do when color printing and adjusting the M and Y filter settings.
One thing that does stand out in the cross-processed strip (immediately above) is that there seems to be a bit of a weird color anomaly in the clouds in the two center frames. Here’s one of them:
The problem is a sort of complex green/magenta crossover that shows up in the clouds in this case. Note the cloud formation in the right half of the frame. This color artifact actually also showed up very strongly on darkroom prints, and no amount of filtering could get it out. In digital space, it also seems to be very challenging to clean up. I think this non-linearity is a tell-tale sign of the ‘unique signature’ (a.k.a. ‘bloody mess’) that this film gives when cross-processed in a C41 developer.
Anyway, to give a bit of a benchmark to this, let’s throw a rather randomly chosen strip of native C41 film (Kodak Ektar), developed in Fuji C41 chemistry at default parameters.
While the film strips above may look sort of normal, give or take a few cc of color filtering optimization here and there, things become more interesting if we look at a positive scan of the actual film strips side by side. Here it is:
Note the differences in the color of the base. While the base of the bottom and middle strips is fairly similar, the top (cross processed) strip shows a lot more base density and a very distinct magenta shift. I think this is further evidence that this film really does not like being processed in C41 chemistry. The difference between the bottom and middle strips in terms of the orange mask isn’t too excessive; it’s not much bigger than what one might see between e.g. certain Kodak and Fuji films, or between fresh and expired film. In fact, I’ve shot quite a bit of very expired Superia 200 that showed a lot more base density!
Another big difference is in the color balance itself. Note how the top two strips are predominantly magenta. Yes, the middle strip is mostly green grass, so it would work out as a lot of magenta anyway, but the top strip with the clouds is also quite magenta, even though there’s little green in most of those scenes.
Indeed, if I look at online examples of amateur shots on Cinestill film, I often see a lot of shift towards green, and it’s sometimes also present in marketing shots by brands offering Vision3 film under their own name. Moreover, in darkroom printing, I actually had to extend the magenta filtering range of my LED enlarger so I could throw a lot more green light onto the negatives to correct for the strong magenta cast in the negatives.
Heck, let’s have a clearer look at this by inverting that side by side scan and color correcting it (approximately) for the Ektar C41 film strip:
Here, the difference in color balance between the strips becomes quite apparent. Firstly, note the strong green cast to the C41-cross processed strip at the top. The increased base density is also visible; I’ve set the blackpoint to approximately the base density of the Ektar strip (which has the lowest density of the three). This also makes a crossover issue in the middle strip visible. While the clouds look slightly magenta and yellow, the film base is decidedly blue, suggesting a blue/yellow crossover as compared to C41 film. This matches my experience from several years ago when I was experimenting with printing ECN2-developed Vision3 film on RA4 paper.
For giggles, let’s correct the comparison above for the middle film strip and see what we get. This is the correction I applied:
This yields the following outcome:
Note how the top strip remains decidedly wonky. The middle one is pretty neutral and usable to my eye, while the Ektar strip evidently now also goes a little funny, although it doesn’t look all that horrible – apart from the tendency of its highlights to go really cyan. I think this illustrates why printing ECN2-developed Vision3 film on RA4 paper is something that’s sort of feasible, with a little bit of luck and given suitable imagery. It’s technically not quite correct (actually, far from it), but I guess that the brain’s signal processing makes us perceive things as normal enough even though there may be some crossover.
Ultimately, I have no choice but to do the final one as well, and correct for the cross-processed top strip, using this curve correction:
Note that the green correction is mostly a non-crossover one: the line remains nearly diagonal. This signifies a green shift instead of a green crossover when compared to true C41 film (which is what we departed from). However, the blue correction is a distinct crossover correction, and I also felt the need to add a red crossover correction in the opposite direction. The following result is obtained:
Evidently, the middle and bottom strips now pretty much go down the drain. They’re also quite dark because I corrected the whole thing for the relatively dense orange mask of the top strip, lopping off shadow detail in the other strips.
Note how the colors in the top strip still don’t quite work out well. This is partly a fluke; e.g. the strong magenta to the clouds in the far right frame are a scanning artefact (the film strip lifted off the glass here). However, the third frame from the left scanned just fine, but it still has an odd orange/red tinge in the sunspot behind the tree. The red crossover correction I applied wasn’t strong enough to catch this one. I tried making that particular correction a little stronger, but this skewed the higher mid-tones, suggesting that it’s not just crossover, not a non-linear one at that. And there’s also that odd green tinge to the clouds, that also suggests some kind of non-linear crossover going on in the magenta/green dimension. It’s a compromised situation, by all accounts. Does it have its creative uses?
All of this is easy in the digital domain – on color RA4 paper, it’s an inherently compromised situation and difficult to resolve. I can sort of see how this might have its artistic/creative applications, although the nature of this kind of crossover isn’t really my cup of tea. I think I’m going to leave cross-processing Vision3 film in C41 for what it is, at least for now, and stick with ECN2 developer instead. It’s not very difficult to make, and my initial printing exercises suggest that usable results can be obtained this way – although a lot more playtime is needed to explore this further.