Unsafe lights – The iffy business of RA4 safelights

Opinions vary on the issue of safelights for RA4 printing. One person says they’re fine, within rather tight boundaries. Others advise against any lighting and recommend working in the dark. I belong to the latter category, and I believe that I have good reasons, too (of course).

It would of course be nice, to be able to work in the not-so-dark when handling RA4 paper. It’s far less painful to cut sheets from a minilab roll than fingers from your left hand because you can actually see what you’re doing. And it’s also convenient to see when exactly you’re going to butt heads with your enlarger. With black & white printing, we can use a dim red or orange/green (for some, not all, papers) safelight. How about color RA4 paper? Color safelights exist, so surely they should work?

I’m a fan of this spectral sensitivity plot of Fuji color paper I’ve shown a million times now, and once again, it comes in so handy:

Notice how there’s an apparent gap in the sensitivity of the paper just shy of 600nm? Well, that gap is the raison d’etre of color RA4 safelights. Theoretically you could use a light source with a wavelength of, let’s say, 590nm and the paper wouldn’t be able to ‘see’ it. That’s the theory at least.

And indeed, it’s practice, too – to an extent. Several safelight solutions exist for RA4 paper. Most of the older ones used an incandescent bulb and heavy band-pass filtration to cut out everything below and above a narrow bandwidth around 590nm. An alternative was low-pressure sodium, which conveniently has a dominant emission slightly above 589nm. At low levels and with some additional filtration to cut out the tiny amount of stray emissions of other wavelengths, this seems to have worked just fine.

In the past.

Because most of these safelight solutions stem from the heydays of darkroom color printing, prior to 2000. But around the turn of the century, the industry went digital and color papers very rapidly changed to suit digital exposure. Part of this change involves an increase in absolute sensitivity of the paper. Color papers of today are significantly faster than they were up to 25 years ago.

But still – there’s an actual gap in the paper’s sensitivity, isn’t there? So even if the paper is faster, shouldn’t it be insensitive to a safelight provided it’s of the correct wavelength?

You’d say so, yes. But the question is whether the gap is really there. And I think (I’m actually pretty darn sure) it isn’t. One reason I don’t believe that gap exists, is because the plot doesn’t make sense. Note how the red sensitivity curve gently slopes off to 600nm and then just stops. I’m not buying it that it really stops there. I think they (Fuji) just stopped plotting data at that point. In reality, the sensitivity of the emulsion will taper off – just like the green plot does on its left slope. This means that there will be some red sensitivity below 600nm, and there may even be a very tiny amount of green sensitivity at that point, too – although I’m willing to believe it’s less than for the red emulsion.

That’s not a weird thought at all. Have a look at the sensitivity plot for (R.I.P.) Kodak Endura:


Endura doesn’t really have a gap just below 600nm. It has a valley at around 580nm, but with sensitivity of both the green and the red sensitive layers. Especially the latter. I have no reason whatsoever to believe it’s much different for Fuji paper.

A second reason why I’m not buying that plot (at least not in this particular case, with this level of granularity) is something I observed recently when I was testing green LEDs for RA4 paper, with wavelengths of 525nm and 550nm. According to the sensitivity plot, the blue emulsion of the paper shouldn’t be able to see either wavelength. Yet, in my testing, it showed very significant response to 525nm light, but not to 550nm light. This tells me that in reality, sensitivity extends past the plot in the Fuji datasheets. If this is true for the data they provide for blue layer, it might apply to other layers just as well.

But hey, we can still try, can’t we? In the past, that’s what I did. I used to have one of those little Durst safelights that had three filters you could rotate in front of a dim incandescent bulb. At some point I gave that a try, setting it up so that I could juuuuust make out what I was doing when handling color paper, at least after having given my eyes the time to accommodate to the very low light level. I was treated to a nasty cyan fog. Hm. Not good.

Especially after that experience, I never really trusted the incandescent lights with filters – filters virtually always have distinct slopes to their transmission plots and that creates problems. They’re simply not perfect.

More perfect, at least in theory, is a single-wavelength light emitter. I already mentioned those LP sodium bulbs, which saw (and in the hands of a few proud owners, still see) use in darkrooms in e.g. the renowned Thomas Duplex safelight units. The modern equivalent would be a 590nm LED, which also has a narrow emission spectrum at a favorable wavelength. This is what e.g. Heiland use in their dual-purpose safelights.

Well, 590nm LEDs are cheap and cheerful, which is why I included a strip of them in the light fixtures I built for my present darkroom. I simply started using them (for color printing!) after some very cursory testing, in accordance with the philosophy “as long as I don’t bump my head, apparently it works”.

My 590nm color ‘safe’ lights. They look hella bright this way, but this exposure was 4 seconds at f/8 at ISO12,800 (EV -3). Still quite bright, but we can still infer one or two things from it.

I bumped my head.

One day I was printing a large series of holiday snapshots and I ran into cyan casts on some of my print borders. Not all around, but usually in a corner somewhere. Investigating a bit further, I also noticed that some of my prints that had clean borders still had odd and variable color casts across the image. Hm. Not good.

And it makes perfect sense, too. Color paper sure is sensitive to 590nm light. It’s just a matter of how much light you throw at it. You may get away with a tiny bit – but how much is a “tiny bit”?

Moreover, silver halide materials, and certainly color paper, respond to cumulative exposure. This means that if you expose the same material several times, the exposures will simply add up (within certain boundaries, which I’ll not go into here). In turn, the implication is that if you expose the paper to a small amount of 590nm ‘safe’ light (i.e. not enough to actually create density by itself), any image-wise exposure will be added on top of it. Since a 590nm exposure turns out to mostly/only affect the cyan-forming emulsion, a ‘safelight’ exposure creates a cyan cast.

Yesterday, when I was printing, I decided to do a little test to demonstrate the problem. It’s quick enough to do anyway. I turned on my (not so) safe lights – see image above, although as experienced in real life, it looks more like this:

This is an approximation of what the safe lights look like in real life – if you peer very intently, you can make out shapes in the room. That’s about the visibility they offer – not much more. The only thing easily visible is the reflected light from the unit against a nearby ceiling straight ahead.

I then placed a strip of paper on the easel (which is one of the darker areas in the room) and made a step wedge with just the safelight exposure:

It’s a poor scan and I should have washed it a little better, but it’s just to illustrate what it looks like. In reality, I can just make out a cyan tone at the 30 second band. Below this, it’s just paper white. So 30 seconds is definitely unsafe. Below that…well, let’s find out.

So I took a small sheet of paper, exposed one half to the safelight in the same spot for 20 seconds. I then exposed the entire sheet with a negative. Here’s that print:

If I look very, very (very!) closely at this print, I think I may see a hint of cyan on the white borders on the left half of the print – as long as I know it might be there. I haven’t tried measuring it with the i1Pro, but I suspect it may just pick it up. However, for all intents and purposes, the borders of this print are white as you’d expect them to be.

The image area itself – not so much. Note how the left side has a very heavy cyan fog. It’s impossible to miss. And this is at a safelight level that doesn’t produce any tone by itself. Ouch…

So how low do you need to go to not just clean whites, but also get rid of the cast? To be honest, I didn’t bother to find out, but what I did do, was make one more test along similar lines as the one above. But this time I didn’t flash part of the paper for 20 seconds with the safelight, but 5 seconds in stead. So two stops less fogging exposure.

In the dark, I messed up a bit and only fogged the left most edge for the full 5 seconds (the other part of the left side, I inadvertently partially shielded; you can still see some color cast, in a gradient running down across the left half), but it still clearly shows the effects of even a much reduced fogging exposure.

So if I take the level of safelight that will not fog the whites, and then go down by two stops, I’m still nowhere near an actual safe level. If I were to hazard a guess, I’d say that around 1 or 1.5 seconds, I might be at a level that wouldn’t be bothersome. Evidently, that’s not of much practical use.

Now, there are many things you could say to counter my argument. But it doesn’t make me any less anxious:

You might argue, 590nm LED is not the same as the very narrow-peaked 589.3nm low pressure sodium vapor, so the old Thomas Duplex lights might be safer. Well, be that as it may, there’s no doubt in my mind that the sensitivity of the paper also extends to 589.3nm – see the Endura sensitivity plot and my argumentation about the (in)validity of the Fuji plot in this regard.

Or you might say, well, your safelight level is just way too darn high – tone it down a lot and it’ll be a different story! Sure, at some point, the problem will go away. Going by my earlier estimate, I could get away with handling the paper for 30 seconds if the safelight is around 5 stops dimmer. That’s EV -8. Problem solved, right? Well, not quite, because an EV -8 safelight is so dim that it’s practically useless. You might as well work in the dark.

About that last point: if you allow your eyes to accommodate to the (lack of) light for 10-15 minutes, even very, very low light levels become visible. This would be the ‘solution’ for working with an EV -8 (or thereabouts) safelight. But in color printing, I find it necessary to iterate between safelight and bright viewing light conditions. After all, I need to judge colors and adjust filtration and exposure, at least periodically. It’s very, very rare that I can remain under safelight conditions for any extended period of time when printing color, and waiting 10 minutes every time before being able to see anything is just a waste of time.

A pragmatist might even argue: so what if the safelight fogs the paper, as long as the whites remain white, just adjust the color filtration on the print to account for the safelight color shift and the results will still be fine. Except that they wouldn’t be, for two reasons: (1) you’d have to accurately control the amount of safelight exposure each sheet and test strip receives, and (2) the gamut of the paper will still be reduced – in other words, you won’t be ever able to get the full range of hues the paper can render under better conditions.

So my conclusion after bumping my head several times over the years, doing some testing here and there and also doing the theoretical gymnastics from time to time, is that a safelight for color work is tricky business. It is possible to use one, but:

  • It needs to be of the right wavelength, around 585-590nm, with little leeway especially downward, because of the risk of activating the green layer, which is much faster than the red layer.
  • You’d either have to use a safelight at such low levels that it takes quite long for the eye to accustom to it before you see anything at all, or…
  • …you can use a somewhat brighter safelight, but only very briefly.

Personally, I choose to simply not bother at all and work in the dark. It’s not all that cumbersome once you get used to it. And it’s absolutely, perfectly safe for the paper.

11 thoughts on “Unsafe lights – The iffy business of RA4 safelights”

  1. Thank you very much for this excellent article that will save me countless hours testing my Heiland safe-light to end up with the same practical conclusion: color printing is safer done in the complete dark even though it takes a little bit to get used to.

    1. Thanks for your comment Régis! This is not to dissuade anyone from using or purchasing a Heiland (or any other) safelight, but I hope it does help in setting up the tests to determine what safelight level is really safe for printing. It’s important that this test not only verifies at which level the white paper turns cyan, but also (more so) at which light level the color balance of the print is affected. I’m sure the Heiland unit also can be useful, but like any safelight, it needs to be used with caution.

  2. You are right, there is no real gap, it’s only a point with lowered sensitivity. This could work fine in the past with some old-school papers, which were less sensitive. Problem is in extreme sensitivity of modern colour papers, made for minilab use, which means that also the sensitivity of the “gap” is much higher than at the old papers. But at least the gap is still retained there.
    For darkroom use we don’t need this insane sensitivity, but is it possible to easily decrease the sensitivity of the paper?
    In this
    In the beginnings of my RA-4 experiments I also tried some amber LEDs and realized that this is balancing on thin ice. I had two shades of amber leds in my drawer – one more orangeish, one more reddish to my eye, maybe 5-10 nm difference? One of them caused cyan and other caused magenta fog in the start. After prolonged exposure both of them caused blue cast as they both hit magenta and cyan layers. With very dim setting I could work relatively fog-free, but it took long till I was able to adapt to such dim light.
    But spectral distribution of such amber LED is quite broad. Of course there is some peak like 595nm or so, but it radiates in some adjacent wavelengths too. It spreads from green to red.
    But the light spectrum can be made narrower by narrowband dichroic filter. For my experiments I bought two pieces of 590nm dichroic glass and put it in front of amber LED. Problem is that the filter works only for lighbeams in axial direction, so I made directional light by putting the LED in short black tube and put the filter on the output of the tube.
    The difference between “fog-less” brightness of naked and filtered LED is whooping. Of course there is still tendency to fogging, but at substantially higher intensity.
    There is still some room for improvement – for example I found that I can fine-tune the wavelength by slight tilting the filter. I can put two filters on each other to make it even narrower.
    It’s clear that making the light more narrow will help only to some point, because even if the light is spot on in the gap, the gap itself is somewhat sensitive. But maybe it will be sufficient for comfortable work with the paper.

    1. Tanks Ivan, great to hear from you again; your response is interesting and thoughtful as always! Yes, LED does have some emissions on either side of the dominant wavelength. I should really set up a test for this and quantify it, because it so often turns out to be a relevant factor. I’m going to think about this. And indeed, no doubt that additional filtration can help make the safe light a little safer still. However, as I was writing this blog, a man in the UK tested his Thomas safelight (low pressure sodium vapor with additional filtration) and observed the same cyan fogging that I see – it just takes a little longer, but this can also be due to the intensity of the safelight, not just its bandwidth.

      I am considering further toning down my own safelights; they’re currently running at the lowest PWM duty cycle so I’d need to intervene at the hardware level to bring it down further (I’m only using 8 bit PWM on these units). For my testing I used all three units in the room; I might simply reprogram the system so only one unit is used, and perhaps put some layer of filtration material over it to further reduce its output. However, since working in the dark is perfectly acceptable to me, I haven’t bothered doing much about the situation so far.

  3. I tried some leds to but had the same problems. It is not usuable. Thinking about a night vision googles.

    1. Arturo, the suggestion of IR goggles is excellent. I should have included it in the blog – and now, you did; thanks for that! Although I’ve never tried this, it seems like the best solution by far for those who want to be able to really see what they’re doing while handling color paper.

      1. Searching on my drawers I found what a I need. An old ccd camera with the IR filter removed, a very powerful IR led and an old 135mm lens to allow to put the lcd camera very close to the eye. It is not gonna be the “beautiful” aesthetically but way cheaper.

        Gonna test if the IR makes something to the RA4 paper. If not, I will put together the parts.

        I will inform you about the process.

        1. Yes, that’s a great idea! The main thing I’d be concerned about is the light from the CCD that might bounce against your face and thus expose the room. The paper should be able to see nothing longer than about 780nm, so as long as your IR LED is above that wavelength, I expect no problem there.

          1. Update. I used it and it works but…it is heavy, more than 500 gr. So, it causes little pain in the neck. I don´t know if I´m gonna use it again. I have cut hundreds of sheets in darkness and I had no problem. So, maybe to do another prototype, i don´t know.

          2. Good news! If you manage to streamline the process, please be sure to post back. The question of how to ‘see in the dark’ pops up from time to time.

Leave a Reply

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