Pinholes in NIR and UVA

Using the E-M1 converted to full spectrum with the Pinhole Pro objectives is possible. Using a 58 mm NIR filter (Hoya R72) attached to the front of the 11 mm Pinhole Pro S11 worked fine, with no increase in vignetting. Using the StraightEdgeU 52 mm or Baader U-filter 2″  with a step-down ring blocked the corners of the image completely. The original 26 mm Pinhole Pro suffers a lot less from vignetting and can be used with these filters of smaller diameter than the front thread of the lens without problem.

Even if pinholes are simply holes, and consequently do not absorb radiation, their very small apertures mean that exposure times need to be very long in the UVA, even in bright light.

The photographs below were taken yesterday between 16:50 and 17:10.

26 mm Pinhole Pro, f:173 60 s, StraightEdgeU filter.
26 mm Pinhole Pro, f:104 0.6 s, Heliopan RG695 (Schott glass) filter.
26 mm Pinhole Pro, f:104 1 s, Heliopan RG780 (Schott glass) filter.

The photographs were white-balanced in PhotoNinja on the clouds. None of them have been converted to monochrome, but using the long pass filter with longer cut-off wavelength resulted in extremely faint false colour.


Pinhole objectives arrived today

Thingyfy Pinhole Pro 26 mm MFT


Pinholes need to be very small to provide a useful image. Consequently the corresponding f-values are small, in most cases f:100 or smaller. This results in either very long exposures, or requires the use of very high ISO values. As we will see in the example images this is less of a problem than what could be expected because as the resolution of the pinhole is low, the images tolerate very strong noise reduction processing without losing there character or mood.

Another consequence of the very small size of the aperture is that dust on the camera sensor assembly becomes much more visible than when using glass lenses and larger apertures. Dust that is not visible at all in normal use becomes very visible when using a pinhole, requiring extensive use the a “heal tool” during file editing.

The positive side of using a small aperture is that depth of field is very deep. Pinholes cannot be focused like glass objectives. In contrast there is an optimal pinhole diameter for maximum resolution, which depends on wavelength (colour) of the light, and focal length. As a consequence of this, macro extension tubes can be used to increase the focal length.

First impressions

The two Thingyfy pinhole objectives arrived today. They seem fairly well built and solid. They have a 58 mm filter thread and the MFT mount at the rear is part of the barrel. The finish seems to be black anodised inside and outside. They fit the camera mount just a bit tight, but nothing to worry about. The 26 mm focal-length Pinhole Pro with multiple pinholes, selected with a ring similar to an aperture ring with click stops, seems to me the most versatile and useful of the two. The range of pinhole sizes is broad (0.80 to 0.10 mm) and allows quite a lot of control over sharpness. It also has pinholes large enough to work nicely with “macro extension tubes” to obtain longer focal lengths. In this case the extension tubes are not used for closer focusing, but instead to increase the focal length.

The wide-angle 11 mm focal length Pinhole Pro S11 has a single pinhole, of a size slightly large for maximum resolution at its focal length, allowing its use with a short extension tube in addition to directly mounted on the camera.

First images

When I came back home from the post office there was still enough light to take a few test photographs of clouds in the sky.

Pinhole Pro S11

With this objective vignetting is very obvious as well as some colour shift towards the edges of the frame. A pinhole will never be very sharp and at this very short focal length (wide-angle) the geometry of the light path results in strong vignetting. Pinholes are used for the “character” of the photographs they produce and vignetting and low resolution are what makes them interesting. Their limitations are specially noticeable in small sensor size cameras like the micro four thirds camera I use.

Pinhole Pro S11 11mm MFT, ISO 1000 f:79 1/15 s.

The Pinhole Pro S11 can be combined with a 10 mm extension tube to form a 21 mm f:150 pinhole.

Pinhole Pro

With a focal length of 26 mm, vignetting is much less than with the wide-angle pinhole. By selecting the pinhole size, it is easy to control the (lack of) resolution to one’s taste.

This slideshow requires JavaScript.

With 16mm + 10mm Kenko DG macro extension tubes focal length is doubled to 52 mm.

10mm + 16mm ext. tubes giving focal length 52mm, pinhole 0.25mm = f:208, ISO 1000, 1/3 s.

This combination works satisfactorily resulting in a moderate teleobjective. It became too dark before I could try using the extension tubes individually. These give 36 mm and 42 mm as focal lengths when combined individually with the 26 mm Pinhole Pro.

AC line frequency and shutter speed

The light output of many types of lamps varies at twice the frequency of mains AC line frequency. Alternating power follows a sinusoidal shape, alternating between positive and negative half cycles. This causes the doubling of the frequency, for each half cycle in the power supply there is pulse in the light output. Two different standard frequencies are used for household electric supply in different regions of the world: e.g. 50 hz in Europe and 60 Hz in USA. To ensure evenly exposed frames one needs to choose a suitable shutter speed: 1/100 s, or a multiple such as 1/50 s in Europe, and 1/120 s or A multiple such as 1/60 s in USA. The figure shows measurements done with an oscilloscope and light sensor, with three different lamps as examples (the LED is not the same as used in later examples, and LED lamps vary quite widely on how stable the light output is).


As can be seen in the figure above, incandescent bulbs, fluorescent lamps driven with electromagnetic ballasts, and even many modern LED-based light bulbs designed to be used as replacements of such bulbs, emit light that is pulsed. How deep are the valleys compared to peaks depends on the type of bulb.

Taking a burst of photographs at f/11, 1/200 s, or 0.5 light cycles per exposure we obtain successive photographs with different exposure. The histograms from nine consecutive frames showing variation in exposure by almost 1 EV.

This slideshow requires JavaScript.

Taking a burst of photographs at f/13, 1/100 s, or 1 light cycle per exposure we obtain successive photographs with identical exposure. The explanation is that if the shutter speed matches the length of one whole cycle, or an exact number of multiple cycles, the exact moment when the exposure starts does not alter the amount of light reaching the sensor. The histograms from nine consecutive frames showing almost no variation in exposure.

This slideshow requires JavaScript.

The phenomenon exemplified above with a digital camera also affects film cameras.

What I describe below, affects only digital cameras.

A further twist related to light sources which have a pulsed output, is the use of the electronic shutter built-in in some camera image sensors. The sensor can be enabled and disabled electronically to adjust the length of exposure while the sensor is continuously exposed to light . In contrast  to most modern mechanical shutters, which open and close extremely fast, this type of shutter does not enable and disable all pixels simultaneously. Consequently the whole image is not captured exactly at the same time. In this case, the fluctuation in the illumination results in banding: some rows of pixels receive more light than other rows during the time they are enabled. Olympus’s E-M1 (mk I) camera had this capability added through a firmware update (as “silent” shutter release modes). Sometimes it is said that this type of shutter cannot be used under artificial illumination. However, the solution to this problem is the same as discussed above: choosing a shutter speed that is a multiple of the frequency of the local AC mains line frequency ensures that all pixels are equally exposed. The next photo gallery shows a set of photographs taken under LED illumination (Osram 6 W, 220-240 V, 2700 K, E-14, driven by 230V AC at 50 Hz) using the “silent” shutter mode of the E-M1 (mk 1) camera. I took photographs of a white wall at each available shutter speed between 1/200 s and 1/8 s in shutter priority mode (edit settings used in Lightroom 6 to enhance visibility of banding: exposure +0.5, contrast +100, Black and White treatment).

This slideshow requires JavaScript.

Least banding occurs at 1/100 s, 1/50 s, 1/20 s, 1/10 s. At shutter speeds slower than 1/20 s, there is slight banding in some cases but barely perceptible. The reason is that once the exposure comprises several whole light pulses, the contribution of an incomplete pulse to the whole exposure becomes smaller as the length of the exposure increases.

A caveat is that modern driver circuits for LEDs not always work at the AC line frequency. Constant current drivers and constant current dimming using well-regulated DC power supplies are alternative technologies that completely avoid light output fluctuations. Specially for dimming, in many cases use of frequencies higher than those of mains power, alleviate the problem. Frequencies can vary widely,  from about 150 Hz, which is enough to make the pulsing invisible to the human eye, to tens kHz in some LED light sources, which is fast enough for photography, except at extremely high shutter speeds.

In the case of fluorescent lamps, high frequency ballasts are becoming common and have in part substituted electromagnetic ballasts. These  ballasts use frequencies in the order of kHz, which combined with some delayed fluorescence of the “phosphors” can almost completely eliminate pulsing of the light output.

Old and new: You may have seen similar banding in photographs taken with film cameras of old TV screens based on CRT tubes. The explanation is that a moving electron beam in such displays also causes flicker, once again at 50 Hz or 60 Hz depending the TV standard used in different countries. In addition, the mechanical shutter of film cameras was based on two curtains, and at high shutter speeds implemented as a narrow window or slit moving in front of the film frame being exposed. See the Wikipedia article on Focal plane shutter for details. Shutters in most digital cameras with interchangeable lenses still work on the same principle.

Histograms were produced with RawDigger and are based on the actual raw data in .ORF raw files. RawDigger is available at

You may be surprised by the plots containing four histograms. Camera sensors do have pixels arranged in groups of four, with one red pixel, two green pixels and one blue pixel. Data for all of them are stored in raw files produced by digital cameras. The two separate green channels are not redundant, they contribute to the spacial resolution of the image.

All illustrations, text and measurements are of my own authorship, and copyrighted.

(c) 2017 Pedro J. Aphalo