Shanghai GP3 100 Reciprocity Failure Correction Tests

Reciprocity is the ability for a light sensitive material so respond equally to different brightness of light, so that a dim light for a long time, or a bright light for a short time will result in equal density of image forming on the film. Reciprocity Failure – technically known as the Schwarzschild effect – is where long and short exposure times low and high brightness light does not result in equal density forming on the film. The problem isn’t the length of the exposure, it’s the failure of the film to respond to the low intensity of light, and exposing for more time doesn’t compensate.

All photographic materials suffer from reciprocity failure, but some materials suffer more than others. General purpose material made for pictorial photography (typically still and motion picture camera film) is designed to have good reciprocity between 1/10,000 and 1/2 second exposure times. Photographic material for printing and enlarging is usually designed for good reciprocity between a couple of seconds and a couple of minutes.

You might be wondering when you’d ever encounter exposures shorter than 1/10,000 of a second: some flash pulses can be as short as 1/30,000 second. Reciprocity correction for short exposures, like flash, can only be made by opening the aperture not extending the time.

Reciprocity failure for long exposures is usually compensated by increasing the exposure time, and tables for reciprocity correction abound on the internet. Unfortunately, reciprocity failure is complex and varies from film type to film type and from batch to batch for any given film. There is no simple adjustment that applies to each and every film and the only method to determine accurate exposure correction is to perform a practical test (that is, expose some film and see how it turns out).

This article is a response to a post I participated in on social media discussing Fomapan 100, a film that is notorious for its reciprocity failure. The original poster was complaining that their pinhole images were overexposed, and they posted a link to a reciprocity correction sheet they had used. I had seen that sheet before, it is a table of exposure times from 1 to 100 seconds in 1 second increments. Exposure is exponential, not linear, so a range of exposure times from 1 to 100 seconds in 1-second increments is not very useful: adding 1 second to a 1 second exposure doubles it: adding 1 second to a 2 second exposure adds only 1/2 a stop, while adding 1 second to a 4 second exposure is an almost negligible increase. The table included the calculation that had been used to generate the corrections, so I used this to generate a table of useful exposure values from 1/2 second to 30 minutes in 1/3 stop increments using the standard set of exposure times used on most light meters (1/2, 1, 2, 4, 8, 15, 30, 60, and so on up to 30 minutes). That table is posted below.

The social media discussion took a turn for the worse when somebody could not believe that the corrected value for 8 minutes was 3 hours and 15 minutes, or a 30 minute exposure required over 16 hours, so clearly I was WRONG.

I decided to perform a practical test and clear up the issue, for myself at least. Unfortunately I did not have Fomapan 100 but I did have Shanghai GP3 100 film, which is another film that has poor reciprocity; it’s not as bad as Fomapan but it’s a close second. I also had a reciprocity correction table for Shanghai GP3 that I could use to perform a test.

The Plan

My plan for the practical test was to:

• make a control exposure at 1/2 second, which is the longest exposure before reciprocity failure begins
• make a series of exposures at 1 second, 2 seconds and so on using the exposure corrections from the reciprocity correction table
• compare the corrected exposures with the 1/2 second control exposure and see whether there is significant difference in density: if the correction table is accurate then all the images should have the same density (exposure brightness).

The simplest way for me to get the different exposures (without using neutral density filters) was to make the 1/2 second exposure with a lens wide open, then close the lens down a stop for each subsequent exposure. I chose a Fujinon W 135mm lens (equivalent to about 35mm full-frame) which has an aperture that ranges from f5.6 to f45 so I can make seven images, from 1/2 second to 30 seconds indicated.

I grabbed my Rittreck View 5×7 camera with 4×5 back and incident light meter (a Gossen Profisix) and walked around looking for a night time scene where the light was 1/2 sec at f5.6. It turned out to be in a reasonably well lit a ferry terminal. The exposures made are in the table below, along with their corrected exposure times. Note that an indicated 30 second exposure requires almost 3 minutes to correct Shanghai GP3 100 for reciprocity failure – over 2 stops correction. Fomapan 100 needs even more correction, over 7 minutes!

To identify each image I used a hole punch to make notches in the edges of the film sheets: the 1/2 second has one notch (the factory notch which indicates the emulsion side of the film) while the 1 second exposure has two notches, and so on.

f-number	Shutter Measured	Shutter Corrected	Notches
f5.6	1/2 sec	1/2s	1
f8	1 sec	2s	2
f11	2 sec	5s	3
f16	4 sec	12s	4
f22	8 sec	30s	5
f32	15 sec	1m 8s	6
f45	30 sec	2m 47s	7

Note that after the first image (1/2 sec f5.6) the camera rotated on the tripod head a small amount when I changed the film holder, so the framing isn’t exactly the same for the first image as the rest. Sorry.

Development

To ensure uniformity, all sheets were developed at the same time in the same tank. Unfortunately I miscalculated the dilution and made the developer half strength (used 16ml of HC-110 instead of 30ml) so the film is under-developed. Surprisingly the images have turned out well enough to draw good conclusions about the exposures; the under-development would highlight under-exposure caused to reciprocity failure more than normal development.

Scanning and Digital Processing

The images were camera-scanned on a light box using a Canon EOS M6 and EF-S 60mm macro lens using the same shutter and aperture setting for each scan. These were inverted – changed from negative to positive – using Adobe Lightroom Classic and Negative Lab Pro (NLP) with the same digital conversion settings applied to all images. Finally, I made a video that compares the negatives to each other on the light box. It’s pretty quick-and-dirty but gets the message across.

Results

The images all have very similar density and shadow detail. Without the edge notches it would be extremely hard to know the exposure each image was given.

The negatives are quite thin, but being night photography that’s to be expected. The well lit area under the roof and the road and pedestrian crossing in the foreground have very good shadow detail. Had I given the film the correct development the shadows may have had bit more density, but the highlight areas would have been even more dense and possible blown out. The reciprocity table I used has a column for recommended development reduction and I’m thinking that this is probably very accurate too.

The bright area inside the terminal and the white painted pedestrian crossing on the road are almost identical between the f5.6 1/2 second image, which did not need reciprocity correction and is the “control” image, and the f45 30 second image which required almost 2 minutes additional exposure correction.

Note that during the f22 8 seconds (actual 30s) exposure a bus drove past and left some light trails. I extended the exposure by 2 seconds to compensate but I don’t think it mattered either way (that is, neither the bus blocking the light, nor the extra 2 seconds made a difference to the overall exposure).

Below is a video comparing the negatives on a light box. The “control” 1/2 second exposure is in the top left: note the other images are almost identical in overall density and contrast, more-so than I was expecting.

Conclusion

From these images I conclude that the Shanghai GP3 100 reciprocity correction table is accurate at least up to the 30 second indicated exposure time, and that reciprocity correction is needed for exposures longer than 1/2 second ie, 1 second indicated exposure and longer.

It would be interesting to test even longer exposures, but doing a similar test with a control shot at 1/2 second would require using neutral density filters and there would be a question of whether the ND filters are accurate enough for such low light levels. To answer this I’d need to test the filters on shorter exposures (before reciprocity failure begins) to make sure any exposure differences are not caused by the filters themselves. On the other hand, the corrections necessary for indicated exposures longer than 30 seconds start to become very long, and I don’t really feel like standing around half the night (or the whole night and into the morning) waiting for the shutter to close.

Long exposures are usually used to capture the motion blur of moving water, traffic, people etc, and for this purpose 30 seconds of actual exposure time is more than enough, which is the corrected time for an 8 second measured exposure. If you want long exposure times then a film like Shanghai GP3 100 or Fomapan 100 that have significant reciprocity failure are a good choice because they won’t require investment in neutral density filters.

While making the reciprocity correction table for Fomapan 100 I was amused by the prospect of a 30 minute indicated exposure requiring over 16 hours. For all practical photographic scenarios that means that the sun has risen or set while the shutter was open.