How to photograph distant objects like man and cars, with resolutions enough to identify face and car numbers? I'm looking for suggestions and cost, 10km range, with good sun light. Thanks.
Please don't take 10km literally. I thought it's the safe distance to do it without getting caught. The video is here for your reference, https://youtu.be/AhLsQPuwQbQ.
You can't. I don't care what you've seen on CSI, this just isn't possible in the real world. Even taking Canon's ridiculously big (and now discontinued. Oh, and $100,000) 1200mm lens, The Digital Picture say:
faces were recognizable at distances up to a mile or more
However, you're talking about six times that distance. You could think about mounting a telescope, but the atmosphere is going to ruin any shots. Even at the "relatively short" focal length of the 1200mm, The Digital Picture found that
Most troublesome for the 1200 L from a distant subject image quality standpoint are atmospheric conditions
and again, you're looking to have six times the problem. Whatever you're trying to do, it's time to find a different plan because this one ain't gonna work.
By specifying that you want to do your surveillance photography in "good sun light", you have already shot yourself in the foot. The best time to do this kind of photography is at night or in the very early morning before the heat from the sun has time to create the "thermals" that make extreme telephoto photography almost impossible, even with the very best equipment.
Assuming that you are happy to work at night (or around dawn) when the "seeing" is best, we need to consider what kind of optics you will require to pull off this feat. If, for the sake of argument, we say that the characters on a car number plate consist of strokes 1cm wide, then we will need to resolve half that width (0.5cm) from your stated range of 10km. This gives an angular resolution of 0.000028 degrees, or 0.1 arc seconds. Conveniently, this is the resolving power of the Hubble Space Telescope.
Hubble's so-called angular resolution — or sharpness — is measured as the smallest angle on the sky that it can resolve (i.e. see sharply). This is 1/10 of an arcsecond (one degree is 3600 arcseconds). If Hubble looked at the Earth — from its orbit of approximately 600 km above the earth’s surface — this would in theory correspond to 0.3 metres or 30 cm. Quite impressive! But Hubble would have to look down through the atmosphere, which would blur the images and make the actual resolution worse.
So, with perfect atmospheric conditions and armed with a boot-leg copy of the HST, your next problem is going to be finding and tracking your target, and more importantly, keeping it in focus. If you've tried to follow wildlife with even decent amateur equipment, you'll know how hard this can be. The HST, by the way, does not have auto-focus.
The standard reference on surveillance photography is Clandestine Photography by Siljander and Juusola. Order it from Amazon if you like but your local security services may take an interest in your purchase.
The general consensus in this thread is that detailed photography of a subject at a range of 10km is exceedingly difficult, and probably impossible using commercially available equipment — and there's plenty of evidence to support that in the other answers.
However, there is a way to photograph extremely distant targets in extreme detail — it's just not commercially available for most private citizens. NASA and other space agencies use this kind of hardware to track launches visually.
Image courtesy NASA, released into the public domain.
This assembly is the long-range ascent-tracking camera, mounted on the Contraves-Goerz Kineto Tracking Mount. It's really more of a telescope, but it does a good job of tracking distant targets in good-enough-for-rocket-scientists detail.
Wikipedia claims that this type of device has a 200-inch (5,080mm) video camera, as well as a 400-inch (10,160mm) film camera. These cameras are operated from Playalinda Beach; the beeline distance from there to LC-39A, the southernmost of the two ex-Space Shuttle launch pads, is 5.923km, however, this camera would be in use later during a launch, when a craft is much further downrange. It's not a stretch to say that it could capture detailed images and footage at 10km.
According to NASA's own website, there are other (FLIR/infrared) cameras on similar mounts with focal lengths between 20 and 150 inches (508mm to 3,810mm), used for medium-range tracking.
Unfortunately, I can't find photos which are tagged as having been taken with either of these devices specifically; searching around generally yields photos of the cameras themselves.
EDIT: This video of the October 2014 Orbital ATK Antares launch failure supposedly has some portions filmed with the long-range ascent-tracking camera.
EDIT 2: Come to think of it, cameras used on military drones may be able to spot fairly fine details at these distances. Pop culture would have you believe that a drone can see facial features of a person from cruising altitude.
Wikipedia claims that a Reaper drone will cruise at 25,000 feet, which is roughly 7.5km AMSL. Assuming Hollywood's assumption is right, and that that the drone isn't always looking straight down, and keeping in mind that its service ceiling is double its regular cruise altitude (50,000 feet AMSL), it's fairly reasonable to assume that the cameras on there can see details at 10km, accounting for turbulence and shimmering, hot air. I'm quite sure that details about the optics on these machines aren't publicly available.
I wouldn't really expect a cutting-edge military drone to be widely available to civilians, though!
As others have said, 10 km is not feasible due to the physics of light and atmospheric distortion. However, I'd like to address another aspect of this that hasn't been mentioned yet: if someone is standing 10 km from you and you're both at the same altitude you won't be able to see them because they'll be behind the horizon!
If a person is 1.8 meters (~6 ft) tall the horizon is ~ 4.8 kilometers away.
Calculated with the formula from https://en.wikipedia.org/wiki/Horizon
distance (kilometers) = 3.57 * sqrt(height (meters))
distance = 3.57 * sqrt(1.8 m)
distance = 3.57 * 1.34
distance = 4.7838 km
In order to take a picture of someone standing 10 kilometers away:
3.57 * sqrt(height) = 10 km
height = (10/3.57)^2
height = (2.80)^2
height = 7.84 m
1 meter = 3.28 ft
height = 7.84 m = 25.72 ft
In other words, you'd need to be standing at least 25 feet above the ground relative to the other person, or they'd need to be standing 25 feet above the ground relative to you, or some compromise between the two of you.
Regardless, your lens won't matter as much as your vantage point relative to the subject of the photo since the curvature of the Earth will get in the way!
Just another thing to consider.
Studies by US universities indicate that a subject can be recognized from a distance of about 45 meters. To recognize a subject 10 kilometers, you need a telescope with sufficient power to cause the subject to appear as if he/she was only 45 meters away. Mathematically, the power of such a telescope must be 222X (10 X 1000 ÷ 45 = 222). Using an instrument at this magnification causes the subject to appear to be 45 meters distant. As an insurance policy, let’s increase the magnification to 250X. Such a lash-up causes the subject to appear to be only 40 meters (10 X 1000 ÷ 250 = 40).
Astronomers are telescope experts. They publish that when a primary lens is used to image photographically, they divide the focal length by 50 to derive the power of the instrument. Using these criteria, if you mount a 50 X 250 = 12,000mm telescopic lens, you could theoretically achieve your goal.
Seems to me, a lens with a focal length of 12,000mm is a scarcity. But wait, our cameras yields a miniature image that must be enlarged; otherwise the images we make are unserviceable. When we view our images on a computer or when we make a print that measures 8 X 12 inches, the software of the computer or the printer applies about 8X magnification if we use a full frame camera and about 12X if we use a compact digital. This magnification applied to make a display image works in our favor. We can reduce the focal length of the telephoto by a factor of 8 or 12, depending on the format used. That works out to 12,000 ÷ 8 = 1,500mm lens for the full frame camera or 12,000 ÷ 12 = 1,000mm for a compact.
My conclusion: To achieve your goal, you must procure a quality telephoto with a focal length equal to or better than the above. Using such a long lens on a moving target like a car is a challenging undertaking. In other words, nearly impossible but maybe you can triumph.
There is this guy called Trevor Paglen. He did a project about photographing classified military bases located in remote parts of the United States. Your question and the video you shared reminded me of his work. He developed a technique called "Limit Telephotography".
From his website: http://www.paglen.com/?l=work&s=limit
"Limit-telephotography involves photographing landscapes that cannot be seen with the unaided eye. The technique employs high powered telescopes whose focal lengths range between 1300mm and 7000mm. At this level of magnification, hidden aspects of the landscape become apparent."
I couldn't find too much about the technique itself but I wanted to share it because it could be helpful. More from his website here:
"Limit-telephotography most closely resembles astrophotography, a technique that astronomers use to photograph objects that might be trillions of miles from Earth. In some ways, however, it is easier to photograph the depths of the solar system than it is to photograph the recesses of the military industrial complex. Between Earth and Jupiter (500 million miles away), for example, there are about five miles of thick, breathable atmosphere. In contrast, there are upwards of forty miles of thick atmosphere between an observer and the sites depicted in this series."
edit: there is a video of the man at work I have just found: Trevor Paglen: Limit Telephotography | ART21 "Exclusive"
Depending on the subject and purpose, having your camera modified for IR may help getting more readable image. IR can cut through haze noticeably better than visible light.
This can be done by a specialized service on many camera models. You would want the IR filter preinstalled so that your camera will become IR only device. Not all lenses play well with IR, some create so called hot spots. You would need to do additional research or test lenses on your own.
You will need very good stabilization of your lens - probably using very good tripod and tripod head.
I think the kind of lenses you might be looking for are actually just medium sized telescopes. Perhaps an 8" Newtonian or a Catadioptric would be good choices.
I've looked at individual trees on far away mountains > 10km and could see the larger branches. They weren't really clear or detailed, but you could distinguish them and they were really FAR. The trouble will be tracking steadily and getting focus at that distance. At those magnifications, little movements translate to much larger ones. You're definitely not going to be doing it while holding the camera while standing or even sitting.
One other idea that I don't think anybody has mentioned yet is using video combined with computer vision software to compensate for atmospheric motion. Adaptive optics are needed for astrophotography in part because the light levels are low. During the day, you could theoretically just shoot a couple of seconds of video at... say 100fps, then use an interframe motion vector analysis algorithm to do partial-frame motion compensation to produce a frame that has higher spatial resolution and lower distortion than any individual frame in the set.
IIRC, this sort of technique has been used to (among other things) undo deliberately blocky video intended to hide facial features. By carefully tracking the subject's motion in the frame and taking advantage of knowledge of the blur algorithm—specifically, researchers were able to determine how the smaller parts of the original image affected the color of the larger blocks in different frames, and then were able to reconstruct a rough unmasked image of the video's subject.
I suspect that the same techniques could be applied to your problem. This approach probably qualifies as even more insane than adaptive optics, but it turns what is otherwise a hard hardware problem into a hard post-processing software problem, which may or may not be better, depending on the situation. :-)
This still assumes that you can put the camera high enough to get line of sight, of course. :-)
Contrary to what most here seem to be saying, you can get unreasonably zoomedin for cheap. If you want to take pictures from really far away,skip "camera" lenses, and go for a telescope. A celestron c90 spotting scope has a focal length of 1700mm and costs about 250$. You need an adapter for a camera, but those are like 20 bucks. That's at the point where atmospheric distortion can be an issue, so angle, temperature, etc plays a part. To take pictures to the horizon will almost always have distortion, but you can make out objects, although they will be small. A car from 10km is quite the chore, but license numbers are in no way feasible. If compactness is less of an issue, then you could go for an orion 180mm maksutov cassegrain (lens width not focal length), as that has a 2700mm focal length. I would avoid other types of telescopes. These are the original "spy" scopes. They were used because they could see very far in low light conditions and were much more rugged and compact than other designs. Treat them like a prime lens that you mount on a tripod and attach a camera to. Don't drop them and that's probably the best you can do without making something custom and ridiculous, this will also push the limits of useful images you can see terrestrially.
You're getting a lot of good advice here. I agree with the answers that say this is impractical — especially regarding atmospheric affects. But you can have some success by substituting sensor resolution for lens resolution.
My incidental experience might be helpful.
I do occasional landscape photography using a 10x12 cm (4"x5") camera. With high-resolution film, drum-scanned at 2,000 samples per centimetre (5,000 samples per inch), I wound up with a quarter of a gigapixel image, which I printed out to 36"x24".
But I couldn't resist pixel-peeping! One shot, in the Tom McCall Wilderness overlooking the Columbia River, incidentally showed cars on the highway across the gorge… probably 5 km away. You could see enough detail on the cars to take a pretty good guess at their make and model!
This was with a Schneider-Kreutnacht Super Angulon 90mm ƒ/8 lens. If it had been shot with the intent of identifying those cars, I should have used my longest focal length, 300mm.
So, to directly answer your question, you don't just go out and buy a lens. It's going to take a combination of equipment, experience, processing, timing, and sheer good luck to even get close to what you want. But if you want to specialize in such things, who knows what the limit is?
What you are looking for is a border security camera system. A combo EO/IR tracking system that can ID and track targets of interest in all weather day/night conditions. Here is one example:
https://www.x20.org/product/m7-ptz-long-range-thermal-imager/
But this isn't really about photography...
Has anyone thought about gigapixel technology? This 320 Gigapixel picture (360 degrees of view) which was taken from the bt tower in london allows you to see individual people to the side of the london eye (about 1.7 miles away), it is not enough to see facial features but you can see a person with a blue jacket and grey/black trousers with a white/light colored back pack walking away from the london eye (to its left)
If you were after a specific spot I think you could possibly get more distance by choosing a different camera/lenses and different range for the photos to be taken in (not 360) But I agree with everyone else the atmosphere will be your enemy before hardware.
