Crossroads – A closer look at Kodak and Fuji digital RA4 crossover

Earlier, I wrote about digital RA4 papers suffering from crossover. I feel it’s time to revisit this topic, because I stumbled across some illustrations that make the issue more tangible. Let’s have a look at some theory and split some hairs!

Bless Kodak and their efforts to publish useful and at the same time accessible data. While writing a recent blog, I came across a 2015 white paper of theirs on silver halide (RA4) paper. I must have overlooked the whole thing, or at least the interesting plots that it contains, which allows a direct comparison between ‘optimized-for-digital’ RA4 paper and old-fashioned ‘optical’ paper. Here are the two plots of interest, side by side:

H/D curves for Kodak silver halide RA4 paper optimized for optical enlargement (left) and digital printing (right). From a Kodak white paper, pp 13-14. Both plots are labeled ‘optical exposure’ because this is the test condition used to make these plots. I’m ignoring the also available digital plots for this blog. My interest is mainly in optical printing, after all.

One of the first thing to notice is that in both sets of curves, the curves shoulder off to different maximum densities. The red-sensitive cyan-forming layer has a significantly higher dmax than the other two, which are spaced more closely. This isn’t particularly relevant because the overall density at this point is so high that it’s virtually impossible for the human eye to discern any color.

More relevant is the crossover in the digital curve. It may not look like much, and certainly not as dramatic as the exaggerated plot I DIY-ed for my previous blog on the topic. But it’s there, sure enough. Note how the R, G and B curves align virtually perfectly for the optical paper, but diverge noticeably for the digital paper. In particular the red channel has a different toe (highlights) and shoulder (shadow) behavior.

The divergence may not look like much, but if you’re trying to print a greyscale and balance it to print a dead-neutral grey midtone, a deep shadow will be significantly cyan. Here’s a simulation that approximates what this would look like:

Simulated effect of ‘for digital’ RA4 paper crossover. On the left a neutral greyscale. On the right, the ‘digital’ paper curve when printed optically. Center, a simulated (using curves adjustment) neutral greyscale as it would print on this paper, showing the effect of crossover.

The example above is slightly exaggerated, since the crossover in the shadow region starts in reality at around 1.3logD, which is already quite dark. A minor color shift at such densities is usually unnoticeable, unless a direct comparison is made such as in the example image above. In fact, the crossover at the toe region, i.e. the highlights, is likely more problematic since the human eye does a pretty good job at discerning hues in light areas. Consider the example image above: to me, the pink/red shift in the very pale highlights look about as noticeable as the green/cyan shift in the deep shadows, while in reality, the quantitative shift in the highlights is far smaller than the one in the shadows.

If we overlay the ‘for digital’ and ‘for optical’ paper curves, we can see another noteworthy difference:

‘For digital’ RA4 paper curve (left) and ‘for optical’ (right)

Notice how the ‘digital’ curves are shifted to the left of the ‘optical’ curves. The amount by which they’re shifted is about 0.3logE on the horizontal axis at a density of 1.0. So in practice, this means that the digital paper is about a stop faster than the optical paper. This change was made mostly because digital exposure relies on exposing each individual pixel separately, and in order to keep printing speed (and productivity) sufficiently high, this means that effective exposure times need to be as brief as microseconds (or less!), while a high-volume optical printer (e.g. an old 1990s-style minilab) would be printing at exposure times of of 10ms up to a second. A digital paper needs to be sensitive enough to build lots of density with very little exposure.

Another difference, but it’s much smaller, is the contrast of the papers, or ‘gamma‘. This is the slope of the H/D curve, particularly the straight portion of it. There’s a bit of a challenge in determining this, because it poses the question what to do with the toe and the shoulder – should they be (partially) included in the gamma calculation, or should we focus on just the straight(ish) part of the curve? Let’s just do a bit of both. Here’s the plot above with the lines drawn in with which we can determine the gamma:

Magenta: gamma-lines (with and without toe/shoulder) for the ‘for optical’ paper; cyan for the ‘for digital’ paper.

The gammas work out as follows:

  • Optical paper, with toe & shoulder: 1.77
  • Optical paper, steep straight part: 3.14
  • Digital paper, with toe & shoulder: 2.07
  • Digital paper, steep straight part: 4.2

As you can see, the gamma is very dependent on how you determine the start and end points, so the figures above may not be suitable for a straight comparison with gamma figures from other sources. As I understand, historically, RA4 chromogenic paper used to have a gamma of between 1.6 and 1.8, and the ‘optical, with toe & shoulder’ figure seems to fall in that bracket. In any case, since the gamma was determined in the same way for both papers above, the comparison does make clear that the digital paper is considerably steeper, i.e. has higher contrast. This, together with the higher absolute speed, meshes with the digital requirement “needs to build lots of density with little exposure”.

If you keep in mind that the exposure scale is logarithmic, all this makes even more sense from a digital printer’s perspective. A paper that is ‘just’ one stop faster and a bit higher in contrast can easily cut power requirements on an exposure system by 60% or more. It means you get the same job done with a laser or LED exposure system that’s less than half as powerful than the optical printer would require. Or you can print twice as much paper surface in the same time…

In the above, I’ve relied on the curves from the Kodak whitepaper I mentioned. The specific paper type isn’t mentioned, but I suspect it is/was Kodak Endura for the ‘optical’ paper and Kodak Endura VC Digital for the ‘digital’ paper. I only use Fuji papers these days, so that leaves me with the question what the situation exactly is in terms of crossover with these papers. I know there is crossover, because, well, Fuji people told me. That’s how I knew this was a ‘thing’ in the first place, because to be frank, I never even noticed it (in fact, I’ve met virtually nobody who did!)

Well, it so happens that I now have a set of curves like the Kodak ones above, for FUJIFILM DPII paper. The problem is, I promised not to publish this set of curves, because, well, they’re not for publication. I don’t think they’re much of a secret; after all, anyone with sufficient dedication/patience (i.e., not me) could plot these curves. What I can do, though, is describe the nature of these curves and how they differ from the Kodak ones discussed above.

First, I’ve determined the gamma for the Fuji DPII paper in the same way I did it for the Kodak plots above. The Fuji figures work out as follows:

  • DPII gamma with toe and shoulder: 1.70
  • DPII gamma only steep, straight part: 3.2

Interestingly, these figures are quite close to the Kodak ‘optical’ paper above (1.77 resp. 3.14), even though DPII is most definitely a ‘digital’ paper (as all Fuji’s papers are).

In terms of crossover, the overall pattern of the DPII paper is the same as the Kodak ‘digital’ curve above, in the sense that:

  • There is mostly red crossover, with the red channel ‘running away’ from the other two channels.
  • Blue and green remain more close together, with blue being steeper than green.

There are also notable, differences, however:

  • The red crossover starts at a lower density. With the DPII paper, the red curve starts to depart from the other two at around 0.7logD, whereas this only happens around 1.3logD in the Kodak example.
  • The Fuji DPII red curve does ‘run away’ like Kodak’s, but slightly less so.
  • On the other hand, the Kodak curve shows a crossover with a depressed red curve at the toe (resulting in a red cast to the highlights), but up to 0.7logD the Fuji DPII curves track perfectly.
  • Fuji’s green curve starts to sag a little in relation to the blue curve at around 1.3logD, whereas in the Kodak paper this only happens around 1.7logD. The shoulder also seems shallower in this paper than in Kodak’s.
  • The dmax of the Fuji paper shows all three curves having virtually the same density, at around 2.4logD. I do wonder, however, to what extent measurements at such high densities are very reliable; after all, 2.4logD is less than 0.4% reflectance. At this point, issues such as flare, glare and any sort of stray light becomes a big challenge in measurements. I also very strongly suspect that the methods for making the measurements of the Kodak curves were different from the method Fuji uses for the curves I received.
  • There is a point where the red curve on DPII shoulders off while the blue curve keeps on rising a little. This happens at around 2.3logD, where the effect is probably invisible to the human eye even under very strong illumination.

I’ve made a similar illustration to the one shown before that shows the effect of the particular Fuji DPII crossover on a neutral grayscale:

Simulated greyscales of Kodak ‘digital’ (center) and Fuji DPII (right)

The left greyscale is neutral. The center one is the Kodak ‘digital’ crossover simulation. The right one is the Fuji DPII simulated crossover effect. Note that this is for illustrative purposes only; this set of greyscales is not intended for a direct quantitative comparison. They are instead meant to give a subjective impression of what a ‘neutral’ greyscale looks like when printed onto these papers, optically.

Objectively, there’s a difference between how the Kodak example crosses over and the Fuji DPII crossover. But they both boil down to green/cyan shadows, and if you were to filter a Kodak print for neutral highlights, there would barely be a noticeable difference in this regard.

There are two areas for which I cannot say much about the differences between the Kodak and Fuji papers. One is absolute speed. I do not have a set of curves that shows the absolute speed of the Kodak emulsions – note that in the chart above, the R, G and B channels are neatly overlayed, which shows that they only show relative sensitivity, also of the three color curves to each other. This sort of chart (i.e. the channels overlayed on top of each other) helps to spot crossover, but it removes any information about absolute differences in sensitivity between the layers, let alone between papers of different manufacturers. And this also means that differences in color balance (not crossover) are not visible, as they would show as the difference in space between the channels for each of these papers.

What I do know from hands-on printing experience is that Kodak Endura paper color balances noticeably different from Fuji DPII, and that Kodak Endura is slightly slower in an absolute sense. Corrected for balance and speed, they look pretty close, however.

Indeed, I’ve got a few prints here that Peter Svensson at AAP-LAB made from a negative of mine, in my presence. They’re on Kodak Premier Endura and a comparison print on Fuji DPII. There is a slight difference in color balance and overall density between these prints because we didn’t bother going all the way to perfectly matching both prints in terms of color balance and exposure. Other than that, the prints are really very similar objectively, and the very subtle differences in crossover detection I could not spot even if I look for it specifically.

A final difference that cannot be read from the H/D curves of papers is color saturation. This is also the main difference between the Kodak and Fuji prints mentioned above. The Endura print does have a slight edge in the saturation and vibrancy of certain yellows. I’m not sure, however, how objective that difference is given the other differences in exposure and color balance between those prints. Sadly, I have no more Endura that’s printable; all I have is so badly expired that its whites have gone a pale green/yellow. Maybe it’ll come back one day. The Carestream plant that manufactured in in Colorado still exists, after all.

Well, that sums it up for now. Back to printing! The previous time I wrote about this topic, some people responded that it surely wasn’t worthwhile optically printing these papers anymore. There’s absolutely no reason for this!

2 thoughts on “Crossroads – A closer look at Kodak and Fuji digital RA4 crossover”

  1. The “secret” Fuji CA DP II curves are here:
    I collected various revisions of DP II data sheet surviving in the internet, and in 2012 datasheet the curves were still there. In later revisions of datasheet the curves disappeared for some reason.
    Are they similar to the curves you obtained from “Fuji people”? According to your description it doesn’t seem so… However these curves are for laser exposure, not for optical. But you got both laser and optical curves from Fuji people, right?

    For me, subjectively, this thing of trying to print optically on “for digital” papers (and other stuff) has some adventurous vibe to it. Having ideal “for analog” paper, chemistry, equipment on hands wouldn’t be probably too appealing for me, as I like rather to take various technical challenges, improvising, searching for workarounds…
    Moreover this crossover issue takes us back to older times of color photography, enabling us to make imperfect old-style photos.
    However my audience is me only, nobody else can see the work behind. Color print is nothing of big significance nowadays.
    In 1990s, as a teenager, I made BW photos a lot, and I wanted to get into color, too. But the informations I had were scarce and obsolete. In our town nobody was selling color chemistry and papers, there were only minilabs, but they were unwilling to resell me some material.
    However the 1990s were in technological peak of color – the films and papers reached their perfection, but I had nobody to introduce me there. There was no internet, no e-shops.
    Now all these things are available to me, but the films and papers are not like in 1990s anymore. One has to search way how to cope with it.
    I must decide if I want rather bleak colors with my old gelatine filters, or if I want to experiment with LEDs to get maybe some more saturated colors. Is it worth it?
    I have tons of Kodak Vision, color paper, all chemicals necessary, but I have also tons of fatigue and lack of time and motivation. Not easy.

    1. Getting started takes courage. Of course, I’d always recommend to just do it! Maybe it’ll give you back some energy by getting going. At the very least, you’ve tried. If you’ve got the materials anyway, there’s very little to lose.

      We live in interesting times; while the future is uncertain, the present gives us the availability of lots of materials and information. Ironically, I think this may be a more interesting time to be working in analog photography than the 1990s, when the technology was at its peak.

      Thanks for posting that old DPII curve. No, it does not perfectly match the curve I obtained from Fuji, although it’s close in a general sense. But since 2012, everything has changed. The paper has changed in subtle ways, and the way they make these curves has also changed. It’s a little hard to tell exactly what’s going on in the curve from the datasheet because its resolution is low. But it shows pretty much the opposite red crossover from what the present curve suggests. Also, in the present curve, the green and blue channels are separated more in the shoulder region; they’re more aligned in the curves you posted. On the other hand, the toe region of the present curve is perfectly aligned across the channels, and the white seems to be more neutral and less blue than in the 2012 version. This may have a lot to do with how the curves are created though, and in particular the measurement and data interpretation. All this has changed dramatically since 2012. This makes the comparison tricky; it’s hard to tell if we’re looking at methodological differences or differences in the actual paper itself.

Leave a Reply

Your email address will not be published. Required fields are marked *