2

My task

I want to use a digital camera to capture slow changes of an object for scientific purposes, subject to the following constraints:

  • The camera is a Canon EOS Rebel T5i.
  • The position of the object, camera, and lighting are fixed. Changes in lighting are negligible.
  • The object is about 20 cm from the camera.
  • My main interest is to capture small changes in brightness or colour of the light diffracted by the object by comparing subsequent pictures.
  • My secondary interest are details of the object’s surface or where the above changes happen, respectively.
  • I do not need to normalise photographs for comparison to references or similar. I only need to compare between photographs taken with the same camera.
  • Changes of the object happen on a timescale of roughly an hour.

Note that this is not for purposes of scientific illustration, but to gather scientific data. I essentially abuse the camera as a photon counter.

What I did so far

  • Use the raw CR2 format from the camera since it has a higher intensity resolution (about 14 bit) than any RGB-based formats (8 bit).
  • Disable all automatisms to avoid them getting in the way of comparing pictures.
  • Set the focus using autofocus on one exemplary object. Then switch to manual focus to freeze it.
  • Use the largest aperture (F2.8) for maximum sharpness.
  • Set ISO to 100 (the camera’s lowest native ISO) to reduce noise.
  • Set the exposure time to the highest value for which the raw data is not capped (except maybe for a few pixels).
  • Ignore other settings such as white balance, as they do not affect the raw data (but are applied in digital postprocessing).

Question

Is this approach reasonable or am I missing something?

My biggest concern is that the resulting exposure time is about ⅛ s, which does not use the full potential of what the context of the scientific experiment allows for, which would easily be 10 s of exposure time.

I know that I could technically combine data from several pictures, but this would be very tedious.

  • 2
    One thing you're missing - using the widest aperture does not give you the maximum sharpness. Most lenses are sharpest when the aperture is stopped down by 1-2 stops. You'll have to experiment with your particular lens to find the sweet spot. What largest aperture does give you is minimal depth of field. – twalberg Aug 6 '18 at 20:58
  • Long exposures introduce noise. So you may want to find the best combo of low ISO/long exposure. For more accuracy, you can 1) take a burst of several shots and average them, 2) take a "dark frame" from time to time and subtract it from the shots. – xenoid Aug 6 '18 at 21:26
  • 1
    @xiota: I fail to make sense of your replacing scientific with creative-historical. I fail to see how either adjective applies to my question and I cannot find any specific meaning of the combination (“creative-historical”) in the context of photography. – Wrzlprmft Aug 6 '18 at 21:52
  • 4
    This question says explicitly "Note that this is not for purposes of scientific illustration, but to gather scientific data. I essentially abuse the camera as a photon counter." Scientific illustration is certainly on topic, but "abusing" cameras for completely non-photographic purposes is as clear-cut as can be. – mattdm Aug 6 '18 at 22:53
  • 2
    This is essentially a specific application of what is more generally covered in the Q&A at manual mode in DSLR with constant iso, shutter speed and aperture, intensity of the image still changes – Michael C Aug 7 '18 at 1:28
4

Is this approach reasonable or am I missing something?

You're making the mistaken assumption that consumer cameras, even those we call "pro-grade", intended for taking creative or documentary photographs are precise enough for what you are asking it to do. The EOS Rebel T5i is not a lab-grade instrument. Neither is the Canon EOS 1D X Mark II, the Nikon D5, the Sony α9, nor any other camera made for taking pictures rather than scientific measurement.

The shot to shot variation of your camera's precision with regard to exposure may well be larger than the frame to frame difference of your subject. Tolerances that are totally acceptable for creative and documentary photography can be woefully inadequate for laboratory measurements.

For more, please see the comments following this answer to Is there a sane reason why ¹⁄₁₂₅ is not, instead, exactly half of ¹⁄₆₀?

Also related:
manual mode in DSLR with constant iso, shutter speed and aperture, intensity of the image still changes, particularly this answer
Why can't ISO, shutter and aperture speeds be controlled more precisely rather than sticking to a fixed scale?

  • Actually, from preliminary tests I know that the camera is precise enough for what I want to do; I just wish to get the most out of it. The reasons why I use a device made for taking pictures are: 1) Taking pictures is still pretty close to what I am doing; I just don’t care about the limits of human vision. 2) There is no cheap device for my purpose that is better than a camera for taking pictures, possibly there even isn’t an expensive one. Remember that a lot of development effort went into making modern cameras the cheap and precise tools they are. – Wrzlprmft Aug 7 '18 at 6:18
  • @Wrzlprmft If the camera has already proved precise enough for your purpose, then what is the problem you are trying to solve with this question? – Michael C Aug 7 '18 at 19:23
  • The sufficiency of measurement precision is not a black and white thing. You never have perfect precision, you just can become less imprecise. The only question is when you expect that improving the precision of a certain step is not worth the effort anymore – but you can never know because a higher precision may reveal new insights (which weren’t new if you knew about them). – Wrzlprmft Aug 7 '18 at 19:52
1

I essentially abuse the camera as a photon counter.

Keep in mind that you are using a consumer grade camera, not a measurement device.

My main interest is to capture small changes in brightness or colour of the light diffracted by the object by comparing subsequent pictures.

[...]

My biggest concern is that the resulting exposure time is about ⅛ s

If you want to measure something, you certainly want to know the uncertainty of the measurement, especially when you are looking for small changes. So when you dial in ⅛ s, is it ±1%? ±10%? Or an absolute offset? Is it negligible?

I'd run some tests how the shutter speed varies and then estimate if it is significant for the expected amount of brightness that you want to measure.

If it is significant, you could include a light source of known fixed brightness in each shot as a reference to normalise the brightness measurements throughout all your images.

  • Why -1? This seems perfectly reasonable. It's like calibrating thermometers in chemistry class. – xiota Aug 7 '18 at 7:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.