wiml@milton.u.washington.edu (William Lewis) (06/08/91)
You know those cheap little autofocus ("point and shoot") cameras
(usually with a sort of sliding thing over he lens): I've been
wondering how the focus mechanism works. I looked at one and the only
things that looksed likely were a pair of lenses behind a plastic
window. Not sonar, then. Do they emit an (IR?) light pulse and time
the reflection (sounds like a lot of fast electronics for a cheap
camera!)? Send a focused beam and measure the square-law-reduced
reflection (wouldn't IR reflection coefficients mess it up)? Or do they
just have a very large depth-of-field and a pair of pretty lenses?
This enquiring mind wants to know ...
--
wiml@milton.acs.washington.edu Seattle, Washington
(William Lewis) | 47 41' 15" N 122 42' 58" W
"Just remember, wherever you go ... you're stuck there."mcovingt@athena.cs.uga.edu (Michael A. Covington) (06/08/91)
I believe they work by comparing the fineness of detail in the image at positions in front of and behind the intended focal plane. Fineness of detail can be measured with just a couple of tiny photocells by determining the difference between the light levels on them. I hope someone who really understands autofocus will come on and post a more thorough explanation. -- ------------------------------------------------------- Michael A. Covington | Artificial Intelligence Programs The University of Georgia | Athens, GA 30602 U.S.A. -------------------------------------------------------
tonya@hpldsla.sid.hp.com (Tony Arnerich) (06/11/91)
They work through triangulation, just like the rangefinder cameras of the 60's. One lens projects a thin vertical stripe of IR onto the subject. The other images onto either a scanning mirror (hooked up to the focus mechanism - this is exactly how the old RF cameras worked), or a diode array. Explanation #2 is just speculation, inspired by the high production volume of diode arrays for AF SLR's. In no case does any imaging take place in the point&shoot's focus innards. tonya@sid.hp.com
wolf@netcom.COM (Buckskin Tech.) (06/12/91)
tonya@hpldsla.sid.hp.com (Tony Arnerich) writes: >In no case does any imaging take place in the point&shoot's focus innards. >tonya@sid.hp.com True of the Point&Shoots, but my Chinon CP9 (as well as many other SLR's) does indeed use imaging to provide autofocus. The image in the viewfinder is projected onto a line array of photodetectors, and comparisons are made to determine whether there's a sharp edge crossing the line. If not, the lens focuses, first one way, then the other, finally getting as sharp a line as possible focused on the array. This leads to several interesting "idosyncracies" in focusing the CP9. First, the subject needs to have an obvious vertical line running through it. If it doesn't, you have to find one, request a focus, then lock it down. Second, the autofocus system doesn't work at all in low light. The CP9 has a high-brightness LED mounted above the lens that kicks in when the autofocus and exposure systems agree it's too dark to focus. It lights up the scene a little, usually enough to get a focus lock. One nice thing about this system is it can be uses along with the manual focus lenses. When using manual focus, you can focus normally with the CP9 letting you know when it sees a focused image. I have astigmatism, and so far, the CP9's been right when I thought it wasn't. I've learned to trust it's opinion. - Wolf
lou@caber.valid.com (Louis K. Scheffer) (06/14/91)
wiml@milton.u.washington.edu (William Lewis) writes: > You know those cheap little autofocus ("point and shoot") cameras >(usually with a sort of sliding thing over he lens): I've been >wondering how the focus mechanism works. I looked at one and the only >things that looksed likely were a pair of lenses behind a plastic >window. In at least some of these models, there are two lenses in the classic rangefinder configuration. By looking at the scene from two slightly different angles (hence the two lenses) you can figure out the distance to the target, at least to the accuracy required by these cameras. On my (waterproof) point and shoot, this causes an interesting side effect- the autofocus does not work under water, since the angles are screwed up by refraction. I think they calculate the angle by doing a correlation of the images cast onto two linear arrays of photo-diodes. The polaroid point and shoot uses ultrasound (or did anyway). This works poorly when shooting through windows. Some fancier autofocus cameras work by maximizing the high frequency content of the image. This works poorly on scenes with low high frequency content, such as clouds. -Lou Scheffer-
ken@sugra.uucp (Kenneth Ng) (06/17/91)
In article <1990029@hpldsla.sid.hp.com>, tonya@hpldsla.sid.hp.com (Tony Arnerich) writes:
: They work through triangulation, just like the rangefinder cameras of
: the 60's.
: In no case does any imaging take place in the point&shoot's focus innards.
: tonya@sid.hp.com
I have a Minolta Maxxum 7000 and I'm quite sure it does not use rangefinder
technology. From much experimentation I have deduced that it works by
maximumizing the contrast between adjacent elements. And apparently
the elements it helps the contrast with are horizontal because it works
best with vertical contrast (ala buildings as opposed to venician blinds).
Modern/Popular Photography had an article on exactly how it worked once,
but I didn't quite understand all of it.
better
: One lens projects a thin vertical stripe of IR onto the subject. The
: other images onto either a scanning mirror (hooked up to the focus
: mechanism - this is exactly how the old RF cameras worked), or a diode
: array. Explanation #2 is just speculation, inspired by the high production
: volume of diode arrays for AF SLR's.
I did see one example of a LED diode array, it was in the NASA Tech Briefs,
an example on a rangefinder with no moving parts. It had an LED array with
each LED pointed in a slightly different angle. BTW: NASA Tech Briefs is
an *EXCELLENT* rag for seeing what can be done with technology.
--
Kenneth Ng
Please reply to ken@hertz.njit.edu until this machine properly recieves mail.
"No problem, here's how you build it" -- R. Barclay, ST: TNG