Pinhole photography
Sections in this page
Photography without any lens is possible with SLR cameras, both film and digital. Simply remove the lens and cover the opening with something containing a small circular hole. The body cap for the camera might be used. This is usually plastic and a pin can be pushed through the centre until it just emerges the other side. It is better though to practice first with a more disposable cover. Cooking foil is good. Paper and card are liable to produce dust that is the SLR camera's enemy (a particular problem for digital users). It is necessary to make it light-tight. For my first attempt I did that by using a 25mm extension tube, forming the foil over that and securing it with an elastic band of the kind dropped regularly by our postman:
Canon S40 7.1mm 1/60s f/2.8 flash
I had several tries at improving on that and my Mark 4 version works best. That consists of a large (0.5 cm) hole drilled in a camera body cap, covering which is a small square of cooking foil taped against the cap on all 4 sides by black insulating tape. A small pinhole has been made in the centre of the foil with a fine sewing needle.
The Mark 4 pinhole: foil behind a body cap hole
The pinhole mounted on a Canon DSLR
An earlier Mark 2 version was a needle hole right through the body cap. That did not work because the hole had considerable depth (about 2 mm) so light rays entering at an angle were obstructed. It is important for the hole to be in very thin material.
Here are some results with the Mark 4 pinhole:
Canon EOS 5D pinhole 0.5s ISO1250 2007:04:02
Canon EOS 5D pinhole 0.5s ISO1250 2007:04:02
The pinhole technique gives really sharp images of dust on the detector chip. This can be a problem with digital SLR's and I will write more about it. For now, here is a detail of the previous picture. Notice the black spots:
Canon EOS 5D pinhole 0.5s ISO1250 2007:04:02
Measuring aperture and focal length
I used GRIP to measure the pinhole diameter as 0.53mm. See how I did that.
Using the same calibrated measurement technique I then measured the distance from the pinhole to the detector chip. That is possible because the camera body has a white mark on the top comprising a short line across a small circle to show the position of the detector inside. The white mark and the front of the body cap can both be seen in this photo, making it possible to measure the horizontal distance between them (47.7mm):
Measuring the focal length - 1
That measurement needs adjusting for the thickness of the body cap because the pinhole lies just behind it. So I made a further measurement using this view (1.6mm):
Measuring the focal length - 2
So the focal length for the pinhole, its distance from the detector chip, is 47.7 - 1.6 = 46.1mm.
The focal length can easily be made longer by using extension tubes. More on that later.
For estimating exposure times we need the aperture as a fraction of the focal length: 46.1 / 0.53 = 87.0. So the aperture expressed as an f-number is f/87.
A possible way of working is:
- to first use the camera's exposure meter while a lens is in place,
- note the settings for ISO, f-number and exposure time,
- replace the lens by the pinhole, and
- adjust exposure time (and possibly ISO) by the requisite number of stops (EV)
* A stop is the traditional photography term for doubling or halving the amount of exposure. In the digital age EV, short for Exposure Value, is commonly used to mean the same thing. For more about stops see my camera techniques page.
Consider an example. Suppose that with a lens mounted and the camera set to ISO800, the meter says you need 1/250s at f/5.6. How many stops (EV) down from f/5.6 is f/87? The following table of f-stops should help (arbitrarily counting stops from 0 at f/1.0).
f-number | 1.0 | 1.4 | 1.8* | 2.8 | 4.0 | 5.6 | 8.0 | 11.0 | 16 | 22 | 32 | 44 | 64 | 88 | 128 | 176 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Stop (EV) | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 |
* To fit the progression this ought to be f/2.0 but f/1.8 is a more common lens design.
The table shows that f/88 (near enough to f/87) is 13 - 5 = 8 stops down from f/5.6. So we will need to increase exposure time or ISO by 8 stops (EV). The exposure time would become 1 second in this example if we leave the ISO at 800.
Using extension tubes
I mentioned that the focal length may easily be increased by using extension tubes. Repeating the calculations above for the possible focal lengths using 12.5mm and 25mm extension tubes (singly or together) gives the following table.
Extension tube, mm | Focal length f, mm | Aperture, f-number | Stops (EV) from f/1.0 |
---|---|---|---|
0 | 46.1 | f/87 | 13 |
12.5 | 58.6 | f/111 | 13.5 |
25.0 | 71.1 | f/134 | 14 |
37.5 | 83.6 | f/158 | 14.5 |
The field of view in each case will be the same as for a lens of the same focal length.
Photographers who are fond of using 10-stop filters to blur their subjects please take note of that last column.
Further considerations
The hole needs to be as small as possible - say half the diameter of a standard sewing pin. That can be achieved by stopping just as the pin punctures the foil. On the other hand the hole must not be so small that diffraction effects start to be significant (the hole must be significantly larger than the wavelength of light but it would be hard to make such a small hole with a pin). We must also consider the brightness trade-off: the smaller the aperture, the longer the exposure will need to be. Then with digital cameras we have to worry about noise. GRIP could help here - add a series of reasonable-length exposures, taken in RAW mode so we have plenty of bits per channel.
The hole must be circular but even then focus will never be sharp. The pixels of a Canon EOS 5D are 8.2 microns (millionths of a metre) across so the 0.53 mm diameter measured above corresponds to 65 pixels on the detector. So we need to scale the image down to fit a PC screen and use some deconvolution to improve the sharpness. We could guess a Gaussian distribution on the detector for each source point. So here's the result of using GRIP to deconvolve the photo:
Canon EOS 5D pinhole 0.5s ISO1250 2007:04:02
Steps used to do that: on the geometry menu scale the image down to 0.25 of its size; on the convolutions menu set a Gaussian kernel of halfwidth 7, a strength factor of 0.3 and 6 passes - deconvolve; this produced a noisy result, so finally use a median filter of half-width 1 (ie, 3 x 3). Now try that on your own pinhole photos!
What, no depth of field?
One nice thing about pinhole photography is that you no longer need to consider depth of field. Objects at all distances are equally well badly focussed. Depth of field is a lens thing. My gnome photo above was deliberately taken at ground level to demonstrate the depth of field.
Why use pinholes on such expensive kit?
My reason is that all such experimentation helps me to understand my camera's capabilities and limitations. I also enjoy the technical challenges in overcoming limitations, even if self-imposed.
Some people find the results more aesthetically pleasing than smooth purist photographs. I know someone who likes this result from my rejected Mark 2 pinhole:
Canon EOS 5D pinhole 1/15s ISO800 2007:03:31 12:05
World Pinhole Photography Day
Takes place each year in April. For more details see www.pinholeday.org. I have submitted my 2007 photo which should appear on their site here.