Multiple short exposures vs a single long one

[BHuij]

In a few places (most recently some writings by Barry Thornton), I've come across the idea that, when making an enlargement, multiple short exposures will yield a different density than a single longer exposure, even if the total number of seconds matches. Or in more concrete terms, if I make a test strip with 5 exposures of 3 seconds each (uncovering as I go), then the darkest strip on there will look different than if I simply expose for 15 seconds, one time.

This makes sense if your enlarger light source has a significant "ramp-up" and/or "cool-off" period instead of just going perfectly from 0% to 100% to 0% brightness when toggled. I know this is true for fluorescent and halogen lights, and likely incandescent as well.

But what about LED lights? My enlarger currently uses a 16x16 array of neopixels as the light source. How "instantaneous" are they? I'm curious whether LED lights have a steep enough output curve when transitioning between on and off to render this a non-issue? I rarely stack exposures anymore anyway as I used localized, stop-based test strips nowadays. But the armchair scientist in me wants to know. Was there some other mechanism by which 3x5s exposures would not = 1x15s?

 

[albada]

LEDs themselves are instantaneous when compared to paper exposure-times. However, if you are turning the power-supply for the LEDs on and off, you are likely to have the same ramp-up/down times with the supply.

Mark

 

[BHuij]

Makes sense. In my case the power supply is constantly on while I'm in the darkroom, since that singular power supply provides power to both the neopixel array and the RP2040 Pico that controls it.

Hoping to have some time in the darkroom tonight, maybe among other things I'll test a fairly extreme case and see how something like a 30x1 exposures compares in final results to 1x30.

 

[Doremus Scudder]

Not only is there the ramp-up/cool-down component to consider. Photographic materials exhibit what is called the intermittency effect, which photo-sensitive emulsions react differently to intermittent exposures than one single continuous one even though the total time is the same.

It is small, but noticeable in film exposures of several short exposures when compared to one longer one. I often use several short exposures in camera to make an exposure to deal with issues of subject movement, passing vehicles or people, etc. I find that I need to give a bit more exposure when doing this if I use more than two or three exposures. I haven't quantified this, but do see it.

Photographic papers should work the same way. There may be some data out there about this somewhere. Maybe one of the more knowledgeable sesitometrists will chime in here.

Best,

Doremus

 

[koraks]

I'm curious whether LED lights have a steep enough output curve when transitioning between on and off to render this a non-issue?

It depends on the LED driver, but in general these will be hard & fast on/off switches because the switching losses would otherwise be unacceptable. This is especially true for LEDs like your integrated RGB LEDs, which have inherently limited thermal dissipation capacity.

More relevant is what @Doremus Scudder indicates above. In addition, reciprocity failure can rear its head if exposures are very short.

I wondered (worried) about this quite a bit when I was working on my LED enlarger light sources. In practice, it turns out to be a non-issue. There are always non-linearities, and they basically drop away against the act of filter (color balance) adjustments you do anyway when printing color. And for B&W, it really doesn't matter much anyway since B&W is kind of resilient and non-critical to begin with.

Btw, the non-linearities can be expected to be on the lower extreme of the PWM range, when pulse widths are particularly short, and with high PWM frequencies. W.r.t. the latter, I think neopixels switch fairly slowly at a few hundred Hz, but you'd have to check the datasheet. Of course, 256 bit PWM resolution easily ends up making very brief pulses indeed even at low-ish frequencies. Let's assume the PWM frequency is around 500Hz; this means that at the extreme lower end of the range (PWM value = 1 out of 255), you're looking at a 7.8us (microsecond) pulse. That's pretty darn brief; equivalent to a 1/127500th second exposure (!) Fortunately, you're virtually never printing with such short pulses since your PWM value will generally over at one or two orders of magnitude higher. Still tiny little pulses, but somehow, but B&W and color papers hold up perfectly fine in my experience. With perfect linearity? Maybe, maybe not - but from a practical viewpoint, there's no problem.

 

[MattKing]

Given the rest of your comments in this thread, I don't expect this is an issue, but just in case you are mixing all sorts of disparate technology ....
The timer you are using could also play a role. For example, if it were an electro-mechanical timer like a Gralab 300.

 

[BHuij]

Yeah. It was more an academic question than anything else. I've been using this DIY LED head for more than a year and have not found it to be any harder to get final results I'm happy with than it was with my old incandescent condenser head (quite the contrary actually since the controller I built for it is way more convenient than my old setup).

As part of the process of designing and coding the system, I actually took into account the fact that the % of nonlinearity in "total light output per second" goes up significantly as brightness values per pixel drop and pulse widths get smaller. I can't remember all of the math, but I did decide on a cutoff value somewhere between 0 and 255 for each pixel brightness value, and got kinda fancy with selectively illuminating various "evenly distributed" subsets of the 16x16 grid in order to allow for a higher degree of effective "neutral density" (really just... less overall light output) before running into significantly nonlinearity.

Got to the point where as I dial in more and more "neutral density" on the controller at various contrast grades, the code base automatically determines in which cases it can simply pass lower brightness values in the G and B channels for each pixel, and in which cases it actually needs to select a pattern of pixels that has fewer total pixels illuminated.

At the end of the day, I estimated I could get at least 5 stops of "neutral density" without introducing either exposure time nonlinearity due to excessively low PWM values, and (hopefully) also without causing problems with uneven illumination across the negative, or uneven distribution of blue and green light, which could theoretically cause weird issues like unintentionally varying contrast in different areas of the final print.

When it was time to stress test, I used a known-good 4x5 negative, made an 8x10 print normally (no brightness reduction, f/11, grade 2, 15ish seconds of exposure, etc. etc.), and then made another copy of the same print, but with the overall light set 5 stops dimmer, and the exposure time doubled 5 times. I developed them together in the tray and found that I couldn't distinguish one print from other in any way after toning, drying, flattening, and examining under a bright light. As far as I'm concerned, it passed the stress test with flying colors, and I've been happily printing with it since (although I don't think I've exceeded 3 stops of "neutral density" at any point since that stress test anyway).

 

[BHuij]

Given the rest of your comments in this thread, I don't expect this is an issue, but just in case you are mixing all sorts of disparate technology ....
The timer you are using could also play a role. For example, if it were an electro-mechanical timer like a Gralab 300.

Certainly would be a variable if I was using an old timer. When I made this LED head, I had to run it via microcontroller anyway, so I figured it made sense to code my own timer with a bunch of luxury functions while I was at it (1/12 stop timer increments, variable contrast and "neutral density" control via rotary encoders, etc. etc.) The controller/timer is 100% custom.

 

[koraks]

Oh yeah, that's smart; I like that approach of selective use of subsets as a way to achieve a 'digital ND'. My LED head designs all rely on monolithic channels, effectively, so the only way to mimic ND was to just use sufficiently high PWM resolution and then use the excess for ND functionality. In this instance it worked to my advantage that I have 16 bits of PWM resolution to work with, and really only need about 12 or so for filtering. This leaves me 4 bits (=4 stops) of ND filtering, so that's where I placed the ND limit in my system. I added that ND functionality in a later version as an afterthought because I ran into shortish exposure times for color prints (far below 1 second), and fortunately it wasn't too difficult to hack it into the existing code.

 

[DREW WILEY]

The more relevant question would be, why do it at all? Multiple exposures potentially complicate the procedure all kinds of ways. Granted, I do have two enlargers which use pulsing additive halogen exposure; but it's monitored and adjusted multiple times per second via a complex feedback loop. Powerful pulsed xenon was once used for UV enlargement exposure of printing plates; but that was some very heavy expensive equipment. LED's perform somewhat differently, but might not like the schizophrenic approach themselves.

Have fun with your experiment.

 

[RalphLambrecht]

Not only is there the ramp-up/cool-down component to consider. Photographic materials exhibit what is called the intermittency effect, which photo-sensitive emulsions react differently to intermittent exposures than one single continuous one even though the total time is the same.

It is small, but noticeable in film exposures of several short exposures when compared to one longer one. I often use several short exposures in camera to make an exposure to deal with issues of subject movement, passing vehicles or people, etc. I find that I need to give a bit more exposure when doing this if I use more than two or three exposures. I haven't quantified this, but do see it.

Photographic papers should work the same way. There may be some data out there about this somewhere. Maybe one of the more knowledgeable sesitometrists will chime in here.

Best,

Doremus

the intermittent effect is an actual phenomenon and needs to be considered; it's always best to execute test strips the same way as the final exposure to avoid unwanted surprises; adding exposures will not yield the same result as a single exposure, even when disregarding the possible fogging issue when adding exposures. When the pauses between exposures are long, a latent image effects may be an additional concern.

 

[koraks]

The more relevant question would be, why do it at all?

You're evidently not aware of how contemporary light sources operate. This refers to PWM, which is inherently a repeated exposure - in fact many many thousands of 'exposures' during a single exposure. It's the default way of dimming LED sources, it's very efficient, easy and turns out to work extremely well also in this application. It also eliminates the need for the power-hungry, failure-prone and complex kind of power management and control circuitry you refer to in the rest of your post.

Have fun with your experiment.

It's the same kind of 'experiment' that's used in multi-$k Heiland units as well as commercial semiconductor-based exposure systems in a wide range of applications (including chromogenic printing). So not really an 'experiment'; just the way we do things these days.

One of the interesting aspects of the analog/film-based photographic community is this tension between conservatism and innovation. As evidenced in this thread, the tendency towards the former often leads to a lack of basic understanding of how things are done in modern times, resulting in comments that miss the mark in an almost comical way.

 

[Nicholas Lindan]

From the Darkroom Automation support files --

There are really no intermittency effects with enlarging paper:

http://www.darkroomautomation.com/support/appnotereciprocityandintermittency.pdf

Lamp warm-up / cool-down, which can appear as intermittency effects, may need to be compensated:

http://www.darkroomautomation.com/support/AppNotePH212LampDelay.pdf

I don't know if there are other timers, besides the Darkroom Automation timer, that have a feature to compensate for warm-up / cool-down and LED light source delays.

 

[DREW WILEY]

There have been a few expensive colorheads with built-in shutters for sake of trimming off both the warmup and die-down phase of the lamp. Or one could use an enlarging lens with shutter, operated by an electronic solenoid. Some process cameras used Compur electronic shutters. My own habit is to avoid exposures shorter than about 10 sec, to minimize any such effect.

And I'm very skeptical of that grossly oversimplified darkroom automation claim that there is no intermittency effect with darkroom papers. Not all paper emulsions are the same anyway.

 

[Nicholas Lindan]

And I'm very skeptical of that grossly oversimplified darkroom automation claim that there is no intermittency effect with darkroom papers.

Interesting - could you please post your experimental results?

 

[DREW WILEY]

I don't trust generic answers. You'd have to run a whole battery of of much more specific tests with a wide variety of papers. But yeah, I have rigged up actual shutters and gosh knows how many densitometer readings relative to the whole wider question. Black and white paper is relatively forgiving; but that doesn't mean every single emulsion or blend of emulsions behaves in the same manner. My critical interest was in how film behaves under an enlarger. I used film for a lot of technical applications in the darkroom. Color printing applications were also critical. One learns things in relation to that which potentially cross over into simpler black and white printing applications. Don't expect me to write a book.

I won't go much beyond that except to state that my own equipment list was a lot more sophisticated than yours. You're offering a device to help people get tighter or more predictable control of their paper densities, and that's to be commended. But actual analytical conclusions aren't so simple.

Even LED options are a moving target. In terms of enlargement, the technology is still rather adolescent, if promising. Let's see what happens. Your test subject paper of MGIV has been in the rear view mirror for quite awhile. And why RC?

No reciprocity failure with b&w enlarging papers? That's quite a stretch.

 

[Doremus Scudder]

There have been a few expensive colorheads with built-in shutters for sake of trimming off both the warmup and die-down phase of the lamp. Or one could use an enlarging lens with shutter, operated by an electronic solenoid. Some process cameras used Compur electronic shutters. My own habit is to avoid exposures shorter than about 10 sec, to minimize any such effect.

And I'm very skeptical of that grossly oversimplified darkroom automation claim that there is no intermittency effect with darkroom papers. Not all paper emulsions are the same anyway.

My solenoid-free shutter for enlarging:

1. Hold card under lens
2. Step on footswitch to turn on enlarger lamp
3. Remove card to begin exposure
(Exposure is long, usually 20-30 seconds and timed with a metronome)
4. Cover lens with card to end exposure

I've never really tested intermittency for prints; I always use continuous exposure for the base exposure. For burning, there is often a pause between base exposure and start of burning, but the burn itself is continuous.

I have observed intermittency effects when exposing film.

Best,

Doremus