multiple exposure on colour film -- order of exposures

I have been merrily doing in-camera multiple exposures on colour slide film for years now, without ever considering the order in which the exposures should be taken, simply assuming unconsciously that it should make no difference. Recently I came across a tutorial that claims the order in which you make the exposures will produce different results:

[from https://thefindlab.com/2017/08/02/shooting-double-exposures-on-film/]

Remember how highlight and shadow detail will be influenced by shooting two frames on top of each other.

-The shadows of your first frame will be filled in by your second frame

-The highlights of your first frame will be mostly lost in the second frame

Here are two photos of the same scenes just shot in different order: in the first image, we shot our model first and the blossoms second. In the second image, we shot the blossoms first and the model second. You can see how the detail in both scenes is influenced by shooting one before the other:

I'm still skeptical, and I wonder if any of the experts here would address this question. If it's true that the order makes a difference then this should be an important consideration when doing multiple exposures, yet I have found no mention of it in any other guide or discussion of the technique, and I have looked through many.....

I cannot speak about double exposure results. Seems I have tried to avoid double exposure ever since one or two ruined shots I was paid to take. That being said:

A valid technique used to increase film speed (ISO) is a pre-flash. We set the shutter speed quite high and use a tiny aperture opening and image, out-of-focus, a uniform target like blue sky, white paper or screen or back-lit opal glass, is imaged. Such a quick (“flash”) exposure alters the film’s exposure threshold. This technique is called “hypersensitization”. Such a “flash” exposure is made just before the picture is taken. It can deliver a substantial speed boost.

I think this is what is happening here. Your first exposure records an image and at the same time, hypersensitizes the film. Thus, the ISO is much higher when the second exposure is made.

This might seem crazy but – during film manufacture the ISO of the film can be altered by a controlled flash exposure using different levels and colors. Additionally, a similar treatment, exposing the film briefly to selected chemical fumes is used. Often a super high-speed film is merely a slower cousin that has been so treated. Thus, the film has been “hypersensitized”.

A “flash” technique is used in lithography along with a “bump” exposure. These are two methods used to control contrast in the process camera. “Process” cameras are giant size devices used to copy artwork and printed pictures. They contain a lamp positioned so that it can fog the film with a brief exposure. The camera is also equipped with a finely ruled screen placed atop the film. The screen fractures an otherwise continuous tone image into a myriad of dots. The result is called a “halftone”. To control contrast a “bump exposure is made, without screen. This technique increases highlight contrast. Conversely, a “flash” exposure with screen is used to decrease contrast.

Allow me to continue --- Another way the film’s response to light is modified is also called “hypersensitization”. Formerly, astronomers used film exclusively when imaging. Films and plates often required a time exposure lasting many hours. Additionally, film undergoing super long exposure times entered into a characteristic we call “reciprocity failure”.

We calculate an exposure based on scene brightness and ISO. Using this method, we arrive at the calculated long exposure time. To our surprise, under-exposure almost always results. This is because film speed plummets when the exposure time is exceptionally long. Astronomers generally work using super long exposure times and they find reciprocity failure provokes even longer exposure times.

Clever physicists discovered they could soak the film in a chamber filled with special gasses. This method of hypersensitization ups the ISO and greatly relieves the effect of reciprocity failure.

Thank you Mr. Marcus for the explanation of pre-exposure and hypersensitization. I should explain my usual method for setting the exposures is the common rule of thumb whereby you multiply the ISO of your film by the number of exposures you are going to do and set the ISO to that on the camera. So if you have 100-speed film and you want to do four exposures on one frame you would set the ISO to 400. I'm not at all sure that the ISO will be higher on the subsequent exposures, from all my experience they are equal.

On looking at the sample photos provided by the authors of the cited tutorial, it occurs to me that the two photos used are not at all the same shots: in the first photo the flower branches are covering less of the model's profile than in the second one, which is taken from a different position. If the shots were exactly the same except taken in a different order, we should still be able to see, however faintly, the fuller field of flowers in the first photo, and there is nothing there.

I understand hypersensitivity as explained, but I suspect that the effect is pretty small for this case.

It says that the effect of two smaller exposures might be more than what one might expect,

but still doesn't depend on the order. In most cases, there will be one exposure at each

spot that is much more than the others.

Only for the case where one part had low, but somewhat uniform exposure, such that it

didn't add texture to the image. Yes the effect is non-linear, but it mostly doesn't depend

on the order.

In contrast, consider some printer technologies, where one might cover over another.

By the way, the hypersensitivity effect is also what complicated safelight testing.

A safelight not light enough to make a visible effect itself, might affect later exposure.

You can see how the detail in both scenes is influenced by shooting one before the other:

Which one of your double exposures do you like best ? Perhaps you should be the only person who is ultimately influenced by the order in which the first and second exposures are shot.

The vividness and clarity of the blossoms within the model in both your pictures are the same vividness and clarity, that's because the model has presented a shadow area and that shadow area seems to be much the same in both pictures, which means the emulsion still had plenty of exposure potential to accommodate near correctly exposed blossoms within her "silhouette". However, the second picture of her is less of a silhouette because it has her blonde hair showing giving her a more personal identity. For this reason, I like the second double exposure better than the first. So taking the blossoms first and model second, was the better way in my view.

If, as you say, you've been shooting multiple exposures for years, then how happy have you been with them, did it really matter in which order you exposed the first and second shot for all your previous multiple exposures?

I'm not sure adjusting the ASA is necessarily the way to go. Watching out for undesirable heavy contrast is what needs to be assessed more than juggling exposure settings IMO

You've said it here ...

The shadows of your first frame will be filled in by your second frame

The highlights of your first frame will be mostly lost in the second frame

I'm sure this mainly determines if multiple exposures are successful or not. For beginners, good starting practice is shooting neon signs at night time, composing one after another in different parts of the frame. The darkness in between each sign is not effected by the subsequent shots, because those parts of the film haven't been exposed long enough to make any difference. Day time is another challenge, but it comes down to how highlights and shadows are worked against one another.

These statements and photos have nothing to do with me, I found them in an online tutorial professing to offer guidance to beginners. The unnamed author proposes that the order in which you take your set of multiple exposures will produce differing results, but I suggest that the two photos presented do not support any such notion--they look different because the shot of the flowers was taken from a different position in the second photo. But I would still like some technical explanation for why the order does or does not matter.

Also, I think both of those sample photos along with the other ones in the original article are rather poor, and probably incorrectly exposed. I am including one of my own below to show that you can have multiple exposures without huge light areas in them. I see no reason to change my exposure compensation method, it works just fine.....

The link didn't work for me the first time but now it is and I got to read it. The author says this at the very bottom and I think it's good advice ...

"Like most film techniques, hands-on experience is the best way to learn, so we encourage you to get out and try shooting some double exposures yourself!"

If you are happy with the results you've been getting, then that's all that matters. The order in which the exposures are taken, and their effect, would come from experience I'd say, with the understanding of how highlights and shadows would be affected.

I have one double exposure to show, it was accidental but I learned from it, the flowers enhance the otherwise dull surrounds, and that's why I keep it. Both shots were the correct exposure and I'm not saying that's right or wrong, but it didn't seem to do any harm to this photo. From my fading memory, the flowers were taken first. I like the idea of double exposures, I'll try and do more of them, intentionally next time.

The article is complete nonsense!

It matters not a jot in which order you combine multiple exposures.

The difference in those example pictures is just a shift in the position of the blossoms, with more sky showing in the silhouetted head in one picture.

Was the article posted on April 1st by any chance?

To get more technical: any given area of film in a double (or multiple) exposure will get some combination of exposures - measured in lux-seconds. Say exposure #1 exposes a certain area to 0.01 lux-seconds, and exposure #2 exposes that same area to 0.03 lux-seconds. The total exposure will be 0.04 lux-seconds, and it doesn't matter if you reverse those exposures; the total exposure will still be 0.04 lux-seconds.

The maths of adding things up tells us that it doesn't matter in what order we add numbers; the total is always the same.

A valid technique used to increase film speed (ISO) is a pre-flash.

That's not hypersensitisation Alan. The simple explanation is that the flashing exposure adds to the main exposure - lifting the shadow detail above the exposure threshold in dark areas, while making very little different to highlights.

Take a flashing exposure of 0.008 lux seconds (about the lower limit for 100 ISO film). That gives us a small density. Then whatever is added to that will show up as a density increase. OTOH highlights receiving an additional 0.1 lux-seconds, will be hardly affected. With a total exposure of 0.108 lux-seconds.

I've never really seen the attraction of multiple random exposures - unless one has a definite design in mind from the outset.

For example, and continuing the bicycle theme: A 'ghost' rider on a bicycle, or a multi-exposure with the rider's legs and the pedals in different positions, as if in motion. Or with a motion-blurred background, as if moving very fast.

The use of flash at night - giving a black background - also opens up possibilites. Maybe that of a phantom cyclist riding the wrong way up a motorway (freeway)!

Otherwise, without a plan, it's just an overlay of random pictures, each one of which, on it's own, would probably be an acceptable picture.... or not! Because once committed to the first (underexposed) picture, it's almost impossible to know exactly what the outcome will be unless the superimposition is exactly mapped and pre-visualised.

Serendipity is one thing, but relying on a happy accident seems to be a bit of a haphazard way of going about things.

"The article is complete nonsense!

It matters not a jot in which order you combine multiple exposures."

Thank you so much Rodeo Joe, that's exactly what I needed to know... all the best.... --Claudio

The article is complete nonsense!

It matters not a jot in which order you combine multiple exposures.

 

(snip)

Well, along with reciprocity failure, there could in theory be cases where it matters.

The first exposure might be so low that it's effect is gone before the second, but not the

other way around.

But in practical cases, you have to be extremely (un)lucky for that to happen.

I suspect also you have to wait a long time in between. And the likelihood decreases

fast with the number of exposures. And in those rare cases where it does actually

happen, it is even less likely that you manage to notice.

Well, since the grain of each piece of film is different, there is really no chance of getting

the exact same exposure twice where it could possibly happen.

More likely an alien from a UFO came along and added to your image.

I suspect that, as Alan notes, in cases where one exposure is perfectly uniform over the film, that it might have a noticeable effect.

If it is not uniform, then the non-uniformity will be larger than the actual effect, enough to cover up for any possible detectable effect.

Even more, the two (or more) exposures would have to be more alike than is, pretty much, possible.

Even more, I suspect that the effect, if any, is not noticeable on slide (reversal) film.

For reversal film, the image is made from grains that were not exposed.

Any small variations in the most sensitive grains are likely to be lost in the reversal process.

I've never really seen the attraction of multiple random exposures - unless one has a definite design in mind from the outset.

Well they wouldn't be "random" if a "definite design" was kept in mind, which would probably follow a certain theme, and the subject matter in each exposure relating to each other.

Though they do appear to be random in cv foto's multiple exposure of the bicycle and piano. I can't make sense of any theme in that one, except possibly "Moving House", items ready to load into the moving house van. Unless of course it was just an experiment in taking a multiple exposure to see how it would turn out.

I'm wondering how many multiple exposures are discarded before the photographer sees one that pleases him/her, and then it will be personal taste, if they've been taken on a random basis.

So a theme is necessary in the first instance IMO, then apply acquired knowledge of film behavior, and skill to execute the theme pictorially.

Not mentioned yet, it seems that it would be easier to get good results, even for shots originally on film,

to combine them as digital images.

After reading all the literature I could get my hands on on the subject over many years; it seems that there is still no definitive theory to explain exactly how film 'grains' react to exposure. And this is literature from respected photo-chemists and researchers working in film manufacturers' labs.

We can see from H&D curves that there is an exposure threshold below which no density is formed by development, but nothing to say how semi-exposed Agx crystals behave below that threshold. For example: do two consecutive exposures of 0.004 lux-seconds render the same density as one exposure of 0.008 lux-seconds, or not?

Does it entirely depend on the length of time between exposures?

If you, as I do, subscribe to the simplistic analogy of film being a 'leaky bucket' that you're trying to fill, then time between low level exposures is obviously critical - i.e. latent-image regression is the leak in the bucket, constantly degrading any exposure and making time between split exposures and time before development critical to low level exposure detail. Hence low-intensity reciprocity failure is easily explained.

So, extending this to multiple exposures, it would make the time between exposures more critical than the order of exposures. However, in most pictorial (non-scientific) use, I doubt that any difference between a couple of seconds delay and a couple of days delay would be readily seen.

After all, people get useable results after leaving film in the camera for weeks (months?) on end before developing the film.

After reading all the literature I could get my hands on on the subject over many years; it seems that there is still no definitive theory to explain exactly how film 'grains' react to exposure. And this is literature from respected photo-chemists and researchers working in film manufacturers' labs.

 

(snip)

 

So, extending this to multiple exposures, it would make the time between exposures more critical than the order of exposures. However, in most pictorial (non-scientific) use, I doubt that any difference between a couple of seconds delay and a couple of days delay would be readily seen.

 

After all, people get useable results after leaving film in the camera for weeks (months?) on end before developing the film.

Longest I have done for a roll of my own is 40 years, for VP126. (And I won't be doing that again.)

Second was 30 years for a roll of Tri-X from my last days in my college dorm.

Not so many years ago, I bought an exposed roll of VP122, for 122 development practice.

It turned out to have a picture of the Mackinac bridge under construction, so from about 1957.

That would have been about 55 years.

Not quite like a leaky bucket, as it leaks slower the more water you put in it.

After reading all the literature I could get my hands on on the subject over many years; it seems that there is still no definitive theory to explain exactly how film 'grains' react to exposure. And this is literature from respected photo-chemists and researchers working in film manufacturers' labs.

I thought it was pretty well understood by now, although this is not my strong point. I'm curious as to how current your literature was... I mean, is it in the past 10, or 20 years, or something on that order?

Pretty well understood still leaves some of the important parts.

Photons generate electron-hole pairs, and they move through the crystal until they find a place to settle.

As well as I know it, and I believe the part that isn't so well understood, there are impurity atoms

(those other than Ag and Br) at low concentration on the surface, which make a place for an

electron to settle. Even more, when it does, it makes a place for others to join it.

The short time reciprocity failure comes when the first one hasn't settled down, before the later

ones come along. And as noted for the leaky bucket analogy, the process is reversible with

only one (or small number) of electrons.

It depends on the exact energy levels for a very small number of atoms.

I don't think the order in which you take your double exposures matter unless you have a specific design in mind. What is important is that you compensate for "over exposure" ! Every time you take an exposure you are cumulatively adding to the original exposure. So depending on the image you have in mind, you might have to use exposure-compensation to decrease the exposure for each frame. For example; for 2 images you might want to decrease the exposure by -1 stop. For 3 images -1.5 stops , for 4 images -2 stops etc.

When I was learning how to use a View Camera(large Format) I kept forgetting to remove the film plate after taking each shot resulting in a double exposure. Of course they were un-wanted double-exposures. Some of those images came out pretty good however, they looked like composites of an entire scene, or event.

Every time you take an exposure you are cumulatively adding to the original exposure.

That's a bit of a generalisation. An against-the-light sillhouette, for example, will add almost nothing in its shadow area. Same as an object taken against a black velvet background. That's the very basis for a lot of in-camera 'trick' photography. But anyway, the OP already said they were compensating for the multiple exposures, by multiplying the film's ISO speed by the number of exposures.

That method accomplishes the same end as applying exposure compensation, and because ISO is linear it doesn't require any Log conversion to or from F-stops. Nor use of a calculator. Mine tells me that 3 exposures should strictly need a compensation of -1.585 stops, which most cameras wont allow to be set accurately.... but then again setting 300 ASA might not be exact, or possible either.

I'm curious as to how current your literature was... I mean, is it in the past 10, or 20 years, or something on that order?

It was fairly up-to-date when I read it!

And I doubt that much research has been done on any aspect of film chemistry over the past 20 years.

I believe that a linear exposure change, multiplying the ISO by the number of exposures, isn't quite right.

As you note in the case of the sillhouette example, some areas will have almost nothing added.

While for many years all we had were averaging (arithmetic mean) light meters, and they do work

amazingly well, better would be geometric mean, or as you note, based on F-stops, zones, or

however you like to measure light on a log scale.

Similar to the way that averaging light meters overweight the highlights and underweight the shadows,

multiple exposure will do the same thing. On the other hand, it isn't so obvious that any other way is better.

(That is, usual scenes don't have a flat histogram on a linear scale.)

While we can't do it with film, it should be possible to digitally add images on a log scale, generating

the geometric mean exposure at each point.

I suspect, though, that we are at least slightly used to the way highlights work in multiple exposures.

(As they do in lighting of non-photographic scenes.) It might be that a geometric mean gives too

much emphasis to shadows.

There is, interestingly, something between the arithmetic mean and geometric mean, maybe too

obviously called the Arithmetic-geometric mean, or AGM:

Arithmetic–geometric mean - Wikipedia

That would all be fine for monochrome, but the OP seems to be doing it in color.

As one color layer might be highlights while another is shadow, and depending somewhat

on human vision, some things might be visible that wouldn't be in monochrome, it is

somewhat less obvious.

I believe that a linear exposure change, multiplying the ISO by the number of exposures, isn't quite right.

Huh! How so?

As previously discussed, light and exposure are simply additive. And whether that's done in linear or logarithmic 'space' doesn't really matter.

If you're talking about the exposure required to get visually equal density variations, then that's an entirely different subject. And irrelevant to multiple exposures.

Yes, F-stop changes are logarithmic to base 2, but 2x2 is still 4. Whether you convert those numbers to their logarithms or not!

Let's plod through it again slowly.

Say you want a total density of 1.0D on the blue line of this film curve family.

That requires an exposure of -2.0 log lux-seconds, which is 0.01 lux-seconds in linear terms.

It really doesn't matter whether that's got by two doses of 0.005 lux-seconds, three doses of 0.003333 lux-seconds, or 4 doses of 0.0025 lux-seconds - all of which are above the exposure threshold of 0.001 lux-seconds for that film.

If we work in log space, that's:

-2.30103 log lux-seconds + 0.30103 (log 2) = -2.0 log lux-seconds

or

-2.47712 log lux-seconds + 0.47712 (log 3) = -2.0 log lux-seconds

or

-2.60206 log lux-seconds + 0.60206 (log 4) = -2.0 log lux-seconds

All of the above answers come out at -2.0 log lux-seconds, which is 0.01 lux seconds in linear terms.

The only difference is that in linear space I could work it all out in my head, while in log space there was smoke pouring from my calculator... figuratively!

So, why on earth would we use a logarithmic base when a linear alternative is available? Such as using linear ISO speeds to get us to the exact same answer that cumbersome F-stop calculations arrive at.