25th March 2004, 06:49 PM
How do camera's know the distance to subject?
there seems to be a lot of hype, does anyone know how its done?
afaik, there's no laser distance measuring or triangulation going on. how do they know?
25th March 2004, 07:14 PM
the lens will "tell" the camera the distance focused..that is usually the subject distance..
25th March 2004, 07:29 PM
Its just a position encoder......
FYI......its basically a position encoder.....which is plated onto the lens barrel. SO for specific discrete position there is a encoded signal/combination obtained from this position encoder and this translate to a distance...."data" which may be send from the lens IC chip to the body.
25th March 2004, 08:19 PM
Originally Posted by sulhan
actually...the camera relies on the CCD sensors located on the camera to focus. Light goes through the lens and is reflected by the sub-mirror to the CCD sensors located below the mirrorbox.once infomation is caculated by the processor the AF drive motor will work the coupler to rotate the focusing lens group.....circuit-board on the lens serve to tell the camera what focal length is used....
26th March 2004, 01:01 AM
doesn't tell me much at all
mathematically how does one derive the distance ? ie: what physics is it using?
considering any scene, to a 2 dimensional sensor, there is no depth information, eg: humans without binocular vision (ie: lazy eye or blind in one eye) have no depth perception. since the lens has only one focal point, is there another hardware that helps triangulates the position ?
or is it complete software based where they GUESS the nearest subject based on contrast and resolution.
26th March 2004, 01:10 AM
Lens to film plane is known, focal length is known, the subject distance can be computed from there.
26th March 2004, 01:12 AM
erm, I maybe wrong but here it goes..
Take a look at your lens. There should be a distance scale on it somewhere, going from the closing min focussing distance of your lens to infinity. There might also be other markings showing depth of field at different apertures or Infra-ed focussing mark, but ignore them for now. Now turn the focusing ring. You will then see the distance changing as you turn the ring: this tells you the distance the lens is focussed at, ie. subject distance.
To relay this distance to the camera, there can be mechanical or electronic linkage which tells the camera the distance the lens is focussing at, as outlined by previous posters.
26th March 2004, 03:51 AM
Zerstorer's got it. There are two common formulas, the first is Gaussian, which I think we covered in secondary school, and it goes like this:
1/focal length = 1/distance to object from lens + 1/distance to image from lens
Newtonian formula is as follows:
(distance to object from lens - focal length) * (distance to object from image - focal length) = focal length ^ 2
These formulas are used by the processors in the lens and camera to derive the distance.
26th March 2004, 06:36 AM
Yes, distance can be computed with the formula. But do SLR actually compute the distance ? What is it for ?
I don't own DSLR so not sure about that.
26th March 2004, 10:24 AM
i'm sorry, i think he *didn't* get it.
Sure, the formula uses 3 variables, 1) focal length, 2) film distance and 3) subject distance. But 2 and 3 are unknown to the camera until focus is established.
YSLee's formula is correct, but how does the camera know the lens-to-film distance WITHOUT focusing in the first place? In other words, the camera needs to establish the subject-to-lens distance (ie focus).
Put another way, we're answering the wrong question. loupgarou's question is how the camera/lens achieves focus in the first place, since the camera's vision is monocular and unable to tri-angulate.
i believe the technology used in SLR AF systems is called 'phase detection' or 'phase difference detection'. Dig it up on the net if you're interested.
Originally Posted by YSLee
26th March 2004, 10:49 AM
I don't know guys but for whatever its worth, in the old days, I am taught that ( and of course it is true) the position of objects etc can make use of the Doppler effect ie wavelengths of light or any electromagnetic waves change when a it approaches/or rebounds off a stationery object, to calculate distance. We can then (in modern terms via a processor) calculated the change in wavelength to get a correspondent distance. That could be why it is so difficult to get a distance calculation when you are shooting animals in cages or close-ups as beam is confused. I remember my Sec4 Physics books ( and I am 50 yrs old!) always use the change in pitch of an approaching/departing train or car honk to illustrate this. But I am sure there are new fangled things available today but use variations of the same principle.
BTW, since I am on the line, can anyone tell me how to get some pix or attachments posted in CS? Thanks
26th March 2004, 11:49 AM
isn't the doppler effect about moving objects? ie: a radar gun uses the doppler effect to calculating the speed of a moving object. ie: you literally need to shoot a beam and time it when it comes back to you.
if I'm not wrong, photographic lenses are passive, with the exception of the AF assist beams/pre flashes, they don't emit anything.
26th March 2004, 11:51 AM
oh another thing, eg; nikon's 3d matrix metering, canon's evaluative metering etc etc, all have the means to locate the nearest object (or at least tries) at each af point.
if you're using a wide angle lens where nearly everything is in focus, how does it tell which af point to trigger?
26th March 2004, 12:07 PM
this is your answer
Originally Posted by ST1100
Originally Posted by VADER
26th March 2004, 12:08 PM
after focus is achieved, this is the 2nd part of the answer
Originally Posted by YSLee
26th March 2004, 12:15 PM
Well, actually, nobody knows how it all works. According to some people, it's alien technology which we exchanged for some human bodies (all volunteers, apparently). There's even a movie about it.
Anyway, some of us tried to reverse engineer the autofocus technology:
Last edited by StreetShooter; 26th March 2004 at 12:17 PM.
26th March 2004, 05:06 PM
27th March 2004, 12:05 AM
27th March 2004, 12:41 AM
Originally Posted by loupgarou
Chris, you are right. Doppler effect works only if there is relative motion between the observer and the "observee" This is why the search and rescue satellites are moving (in low Earth orbit) so that they are able to estimate the location of a rescue beacon by Doppler effect. Geostationary satellites cannot do this.
27th March 2004, 05:27 AM
Erm, I'm not too sure whether my explanation is correct or not, so don't worry about it It's more of a guess based on practical observations rather than technical knowledge...
Originally Posted by TME
If you have a close-focussing or macro lens, it's easier to see that the point of focus of the lens is related to the distance to the object. Try focussing at a spot. Now move in and out: the image goes from blurred to sharp to blur again. The point of sharpest focus is your lens to subject distance and also the focussing distance stated on your lens. A chip in the lens (or something) then relays this focussing distance to the camera. When autofocussing, the autofocus system determines when the image is sharp.
OK.. now waiting for someone to tell me that my guess above is total bulls**t