How does a camera remember focus defect of different lenses?

[adakshi]

Hello All,

While discussing the depth of field and depth of focus in other topic, this question came to my mind .. rather several questions.

I know except a few copy probably all consumer lenses either back/front focus to different degree. I read somewhere that there is a screw in the mirror box of the camera that can be turned to nullify the defect. Moreover, some modern cameras provide UI to do that and also remembers the setting.

1. How a turn of a screw can resolve back/front focus issue of a lens? As I understand, if the image plane doesn't coincide with sensor plane, front/back focus occurs. by turning a screw do we move the sensor plane? But then it will back/front focus with other correct lenses.

2. Or do we, by turning a screw, communicate with focus motor and tell it to rotate more/less for every focal distance? But does the amount(cm/inch) of back/front focus remain the same at all focal length? Also, doesn't it mess up with viewfinder? I mean if before adjustment, sharply focused object in VF appears back focused when shot, wouldn't it appear front focused in VF while it is actually correctly focused when shot?

3. What is the relation between depth of focus on depth of field?

4. Wouldn't amount of front/back focus depend on color of light? In that case how a modern camera can use the focus correction info from its memory and apply?

Ok .. I should stop here .. there are a lot questions that keep on bubbling in my mind . I'd be glad if you can help me with the answers.
I am an electronics engineer working in IT and holding a bachelor in Physics hiccup! , so, you know I am trying too many things at a time, I'll definitely dig deep into this DSLR technology, provided I get some time

Now you know why I like Sony!!! Invention !!

Any help is appreciated.

Thanks
Ayan(Sonian)

[David Kilpatrick]

Ayan, I'm surprised you've got a Physics degree and don't know optics backwards! I guess it's not that important now. I failed physics at O-level (15) because of maths - I am disnumerate with written figures - but by that time we had already traced lightpaths, built compound lenses, used the optical bench to check theory against real conjugate distances etc! So stuff like virtual image, real image, focal plane, apertures, ray tracing was driven in long before I dropped out of sciences into arts. Mind you that was in the 1960s and basic stuff like optics was treated as important, most of today's physics didn't even exist, even at secondary school level.

The screw adjusts the position of the focus sensor, which is supposed to be in a plane exactly corresponding to the focal plane, receiving image forming rays via a semi-silvered section of the reflex mirror and two further mirror/prism stages plus a lens. Often enough, it's not precisely collimated to the focal plane. Because it costs real money to make physical adjustments and get two mechanical elements within the tolerance needed, the camera makers instead measure the error on an automated test bench, and create a lookup table based typically on a 50mm lens focused at 2 metres. This table is then used to 'nudge' the focus point every time, after focus is confirmed. But it can be wrong for lenses of other focal lengths, notably very short ones like 10mm; and transport/use can shift the position of the components. Adjusting the screw does not reset the lookup table to zero, so whatever you do, something will always be inaccurate.

Yes, colour of light does affect results, Minolta/Sony used a calibrated source supposedly round 5800K with minimal spectrum spikes. The cameras will generally backfocus in tungsten light or when aimed at red targets, and front focus if aimed at a blue pattern (or used in blue light, which is not common).

Digital is very demanding about focus because the sensor forms a single plane. Film often has up to 14 turbid translucent layers and focus can only hope to be 'within' the emulsion. If you miss the surface, you may hit a bit deeper in. Digital, if you miss you miss. The sensor gets circles of confusion (slices across cones of rays, and therefore not actually circles towards the edges but ellipses with a tendency to coma - that sort of comet-tail like figure of eight shape you see if you focus the sun with a magnifier at an angle). No lens ever really focuses light to a perfect point but digital sensors love those which get close.

David