2
\$\begingroup\$

I am attempting to gather data for creative historical purposes. However, I'm a computer scientist, not a physicist or a photographer, and there seem to be a gazillion considerations.

The environment is the lower deck of an airport loading zone with low light. I need to be able to recognize ("resolve") a person at 30m to 40m in a box of at least 30x60 pixels at 1x magnification. Ideally, the depth of focus will be about 20m to 60m.

With these qualifications in mind, what metrics should I be interested in when purchasing a camera and/or lens?

Besides depth-of-field, resolution, and lighting, are there any considerations that I might add, or ought to be considering, that I'm totally missing?

\$\endgroup\$
21
  • 1
    \$\begingroup\$ Hi Scott, Welcome to photography stack exchange. I have removed the video references to comply with the subject of photography which will still give you what you need to know about the issue. \$\endgroup\$
    – Stan
    Commented Jul 23, 2018 at 18:09
  • 2
    \$\begingroup\$ I'm voting to close this question as off-topic because it is about using a camera to make measurements for a purpose other than producing images for creative or historical purposes. \$\endgroup\$
    – Michael C
    Commented Jul 24, 2018 at 8:07
  • 2
    \$\begingroup\$ @MichaelClark, the help page does not explicitly state that only creative or historical motives are allowable. In any case, the motives are utterly irrelevant to the question, so I'll make something up to make you feel better :). \$\endgroup\$
    – Him
    Commented Jul 24, 2018 at 13:43
  • 1
    \$\begingroup\$ Hi Scott! You noted in meta that you are interested in the photographic aspect here, although for scientific and not "creative or historical" purposes. The problem isn't motives, but what the question itself is about. It's often the case that these end up being "XY" problems, where photography isn't really part of the answer at all. But let's see what we can do here. :) I have a couple of immediate questions... \$\endgroup\$
    – mattdm
    Commented Jul 25, 2018 at 8:22
  • 2
    \$\begingroup\$ then you should provide "i think XYZ because of WQE. Is it reasonable?" \$\endgroup\$ Commented Aug 3, 2018 at 6:03

2 Answers 2

2
\$\begingroup\$

Starting with a desired image height of 60 pixels on a sensor, we need to start with a particular sensor in mind - using my 7D Mark II as an example, the sensor is 15mm in height and produces an image 3648 pixels high - that's 243.2pixels/mm, which means your 60 pixel image covers about .25mm on the sensor, or about 1.67% of the total image height. That means that, in order for, say, a 1.8m tall human at 30-40m to map to that height on the sensor, the total vertical field of view will be about 108m, which at 30m distance implies an angle of view of about 74.5 degrees (or 69.7 degrees at 40m distance). That's the equivalent of about a 10mm lens for the 30m distance, or an 11mm lens for the 40m distance. Those are very wide-angle lenses. So, the bottom line is that you're going to have a hard time getting the images of the people that small at that distance. A more normal lens of say 24-50mm will give a bigger image of the person than you were asking for. If that's a problem, you can always scale the image down in post-processing, or use a lower-resolution mode of the camera, or something.

As far as depth of field goes (there are many online calculators or mobile apps out there for this), to get a 20m depth of field at 30m with a 50mm lens requires about f/1.4; at 40m it requires only f/1.0. A 60m depth of field would be either f/2.8 or f/1.8, respectively. So, even relatively slow f/4-5.6 lenses will give you plenty of depth of field.

\$\endgroup\$
6
  • \$\begingroup\$ Apologies. By 30x60 box, I mean at least 30x60, because otherwise the humans are not detectable by our software. I will edit my question to reflect this. This suggests, I think, that your 7D Mark II would have a 30x60 human at a much further distance? Maybe 80m or so? \$\endgroup\$
    – Him
    Commented Jul 23, 2018 at 19:49
  • \$\begingroup\$ Also, it seems from your answer that aperture is what provides me with depth of field. However, aperture also relates to my lighting problem. Is it correct that contrast and depth of field are a trade-off? What is the relationship here? \$\endgroup\$
    – Him
    Commented Jul 24, 2018 at 16:31
  • 1
    \$\begingroup\$ @Scott Depth of field is determined by 1) aperture, 2) distance to subject, 3) focal length of lens and 4) final presentation of image (i.e. viewed on a screen from half a meter away, or printed on a wall and viewed from 50 meters, or ...). Contrast is more a function of having sufficient lighting, and by noise introduced by ISO (sensor amplification), assuming the scene has sufficient contrast to begin with. \$\endgroup\$
    – twalberg
    Commented Jul 24, 2018 at 16:41
  • \$\begingroup\$ @twalberg DoF is determined by two things: aperture and magnification. Total magnification is a result of focal length + subject distance + sensor size + enlargement ratio needed to display at a specific size + viewing distance. All of that determines the angular size as seen by the viewer. \$\endgroup\$
    – Michael C
    Commented Jul 25, 2018 at 0:34
  • 2
    \$\begingroup\$ Note that depth of field is a perceptual thing — the lens can still only focus in one plane. There is a certain threshold where everything is in focus to the extent of the resolution of the systems, and another threshold where everything is "in focus enough" for a given observer in particular conditions. \$\endgroup\$
    – mattdm
    Commented Jul 25, 2018 at 8:38
1
\$\begingroup\$

I'd like to focus on your depth of field problems and how that relates to ISO. If you're only concerned about one "depth," per se, or distance from the camera, this will not be a problem. The reason for this is that no matter how blurry the background is (or isn't), your subject will always be in focus. However, if you're interested in also picking up subjects closer than that distance, you'll want to make sure that you don't have too low of an f/stop number (too wide of an aperture). If you're using a small sensor, or a wide angle lens, this should not be a problem if you keep the number above f/1.4 or so. However, if you choose to use a telephoto lens to zoom further, you could run into problems with a narrower depth of field.

I'm assuming though, since you're collecting these photos for scientific/historical/data purposes, you'll be less concerned about having the "blurry background" bokeh effect, and more concerned about having everything as sharp as possible. This is where ISO comes in. To achieve a higher depth of field, with more in focus, you'll have to use a smaller aperture (higher f/stop number). That means there's a smaller hole in the lens for light to come through, and you'll need more digital amplification, or ISO. Newer cameras generally do better with ISO than older ones by a good deal, as this technology has been improving dramatically in the last few years. A quick google search with any DSLR name and "ISO performance" will give you information on how that camera performs when it has to amplify the signal. Note that as ISO goes up, noise gets amplified as well as signal, hence the need to check the performance out beforehand.

That said, for your purposes, a used Nikon D3100 with a cheap lens would probably do the job. (I'm not familiar with Canon, so I can't speak to their lineup.)

Finally though, a distance of 90-120 feet is not outrageous by any means. I don't know exactly what you mean by "lowish light," and I don't know how recognizable you need the people to be, but if it's not pitch black, and you don't need to do facial recognition, almost any modern DSLR could get you what you need there. Your next consideration might be price point (and whether you can use the same sensor for video, although that's off topic here).

Considering price point: Look at phone cameras. For example, a used iPhone SE can be had as of writing on eBay for around $80, and seems to do fairly well in other than pitch black. One bonus here is that the sensor size is so small that you don't need to worry about depth of field. For any aperture that the iPhone can do, anything past 6 feet or so will be in focus.

So in short:

  1. Check the resolution. Under about 10 MP you won't be happy with. 12MP is better. For either of these, just make sure you get a 35mm-adjusted field of view of around 24mm or higher. This is still a pretty darn wide lens. 20MP or higher is a bit overkill for your requirements, but if you go that way, you can go with 10mm or so (although these are usually special lenses that you pay extra for, they're so wide).

  2. What depth of field do you need, and therefore what aperture and ISO should you be looking for? If you go with a DSLR, just get your run-of-the-mill cheap lens, and you'll be set. Fairly small aperture (f/4 or higher), and a good field of view for what you're doing (24-55mm should be about right). Don't worry about this for phone cameras.

  3. Look at cheap options. For what you're doing you don't need a $30000 camera. You don't even need a $3000 camera. You may not even need a $300 camera. Depending on whether you need to do facial recognition, and how dark exactly it is, something as cheap as an old iPhone could do the job perfectly. (Bonus: video capabilities).

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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