Learning more about photosensors in cameras

At 9:23 -0800 2004-11-09, ChristopherSchmidtZone.com wrote:
>I also have problems with the latency between my clicking the
>shutter and the camera's latching of the image. I'm not sure if it
>is a sensor issue, or other electronics in the camera. It is very
>frustrating. (I realize that writing to CF will introduce delay
>*after* latching the picture, but I could live with that.)

On my camera, most of that delay is the auto-focus time. You can ask
the camera to stay in focus all the time, but that takes power and
decreases the lifetime of a charge. As a compromise, my camera
allows you to select and lock a focus (and exposure) by
"half-clicking" and holding the shutter release.

My camera (Canon G1) is now several generations old, so there have
undoubtedly been advances, but the need for physical motion to adjust
focus pretty much guarantees that it takes time and power.

--Mark Williamson
=mark.williamson

On Nov 9, 2004, at 12:23 PM, ChristopherSchmidtZone.com wrote:

> Is there a good web site that compares sensor size, "film speed",
> latch latency, and minimum shot interval for low end cameras (under
> $400)? I've found such data only sparse and sporadically available,
> though it would seem like the kind of thing testing labs would measure
> routinely and compare.

My favourite for Digital Camera reviews is:

http://www.dpreview.com/reviews/

Their reviews are quite detailed, and they sometimes do comparisons
between cameras in similar categories.

Roger
Roger Henriques
rdh at rhen dot com

On Nov 9, 2004, at 10:23 AM, ChristopherSchmidtZone.com wrote:

> Is there a good web site that compares sensor size, "film speed",
> latch latency, and minimum shot interval for low end cameras (under
> $400)? I've found such data only sparse and sporadically available,
> though it would seem like the kind of thing testing labs would measure
> routinely and compare.

Imaging resource has excellent and detailed reviews of many digicams,
including some of the metrics you're looking for.

<http://www.Imaging-Resource.com>

--Nik

On 09-11-2004 18:23:41, ChristopherSchmidtZone.com wrote:

...

>Is a larger sensor like using faster film--or slower? Do you need a bigger
>lens area to gather more light to feed a larger sensor? Or do modern CCD's not
>need a lot of photons to work well?

From my limited experience, CCDs need as much light as film does to work well. On my
camera, I can select the CCD's sensitivity, but at full resolution it doesn't go any
higher than 400 ISO.
This is one area where digital cameras could use a lot of improvement: a camera that
doesn't need a flash to make good indoor photos would be so much easier to use than what
we have now. To do that, you'd need a sensitivity that's equivalent to 10,000 ISO...

>Is there a good web site that compares sensor size, "film speed", latch
>latency, and minimum shot interval for low end cameras (under $400)? I've
>found such data only sparse and sporadically available, though it would seem
>like the kind of thing testing labs would measure routinely and compare.

<http://www.dpreview.com/> has lots of information.

Harro de Jong

Mark Williamson wrote:

On my camera, most of that delay is the auto-focus time. You can ask the camera to stay in focus all the time, but that takes power and decreases the lifetime of a charge. As a compromise, my camera allows you to select and lock a focus (and exposure) by "half-clicking" and holding the shutter release.

My camera (Canon G1) is now several generations old, so there have undoubtedly been advances, but the need for physical motion to adjust focus pretty much guarantees that it takes time and power.

I was always under the impression that autofocussing mechanisms needed "edges," not movement, in order to bring the subject into focus. That's why some cameras have poor autofocus mechanisms in low light and others use a type of preflash lighting to overcome this problem.

With both my Sigma SD-9 and Canon G2 they won't come into focus if the focus area does not have an edge for them to work on.

John Wolff Hamilton, NZ

--- begin forwarded text

From: "Charles Maurer"

>Christopher Schmidt asked:
>Is a larger sensor like using faster film--or slower?

Both and neither. Larger sensors are like slower films in that that they
show less noise. Larger sensors are like faster films in that they can
take pictures in less light--but only because their lower noise allows
their output to be amplified more. For optimal results, the sensitivity
of all image sensors seems to be more or less the same.

>Do you need a bigger lens area to gather more light
>to feed a larger sensor?"

Yes. If the sensor is larger, then to cover the sensor, the lens must be
longer in proportion. If the lens is longer, then to cast the same
amount of light on the sensor, the aperture must be proportionately
broader. For this reason, f/4 in a 20mm lens is 5mm across (20mm/4) and
f/4 in a 40mm lens is 10mm across (40mm/4).

This leads to the one optical disadvantage of a larger sensor: depth of
field is proportional to the diameter of the aperture, so with a larger
sensor, comparable depth of field requires a smaller f-stop and longer
exposure.

--

Another reader asked me to explain how the size and resolution of a
sensor interact with cropping. This would be a sensible place to discuss it.

Cropping a photo throws away information. For this reason, to make a
smooth enlargement, more pixels need to be interpolated.

The eye sees a cropped photo like any other photo. In almost any photo
used for pictorial purposes, the contrast of edges will matter more to
the eye than resolution. Cropping does not affect this. Indeed, the
examples on <http://www.tidbits.com/resources/751/Sharpness.jpg> are the
equivalent of croppings from enormous enlargements.

Enlarging a picture magnifies blurred edges and noise. The eye will
notice blurred edges or an overlay of noise before it notices that some
minuscule detail is missing. I find that enlargements from my pocket-
sized camera are usually limited by noise, and enlargements from my SD-10
are usually limited by the sharpness of edges. Merely squeezing more
cells onto a sensor may sharpen edges a mite (or it may not: the optics
of this are not straightforward) but it is almost certain to increase
noise. Thus, squeezing more cells onto a sensor may allow more cropping
or it may allow less.

In any case, squeezing more cells onto a sensor will reduce its dynamic
range and for pictorial photography, reducing dynamic range loses more
information than reducing resolution. These photos demonstrate this:

http://www.tidbits.com/resources/755/MonkRaw.jpg
http://www.tidbits.com/resources/755/MonkAdjusted.jpg

I made a 6.75" x 10" print of this picture with my dye-sublimation
printer. If I look at the print with a magnifier, I can see the
individual hairs on the monk's head. Obviously the screen prints above
have lost a lot of resolution. However, it is also obvious that even if
you could get close enough to your monitor to make out the hairs on the
darker picture, the lighter picture as it stands would still be better.

--- end forwarded text

At 01:39 PM 11/12/2004 -0800, dtopcomp wrote:
>Mark Williamson wrote:
>>the need for physical motion to adjust focus pretty much guarantees that
>>it takes time and power.
>
>I was always under the impression that autofocussing mechanisms needed
>"edges," not movement, in order to bring the subject into focus.

Yes, but the way it detects edges is by moving the lens and finding the
lens position which results in the sharpest edges. In difficult-to-focus
situations, I can hear the lens moving for a couple of seconds before the
camera either focuses or tells me it can't figure it out.

Thus Mark's comment refers to motion of the lens, not motion of the
subject. And moving the lens takes time and power -- a lesson I learned
early, when I left the camera in a "full time autofocus" mode and ran down
the batteries after only a few dozen shots.

For cameras which have a manual focus option, going to manual focus can
dramatically reduce the latency. I can detect no latency at all in manual
focus mode on my camera.

Edward

On Sunday, November 28, 2004, at 09:03 AM, Edward Reid wrote:

> At 01:39 PM 11/12/2004 -0800, dtopcomp wrote:
>> Mark Williamson wrote:
>>> the need for physical motion to adjust focus pretty much guarantees
>>> that
>>> it takes time and power.
>>
>> I was always under the impression that autofocussing mechanisms needed
>> "edges," not movement, in order to bring the subject into focus.
>
> Yes, but the way it detects edges is by moving the lens and finding the
> lens position which results in the sharpest edges. In
> difficult-to-focus
> situations, I can hear the lens moving for a couple of seconds before
> the
> camera either focuses or tells me it can't figure it out.
>
> Thus Mark's comment refers to motion of the lens, not motion of the
> subject. And moving the lens takes time and power -- a lesson I learned
> early, when I left the camera in a "full time autofocus" mode and ran
> down
> the batteries after only a few dozen shots.
>
> For cameras which have a manual focus option, going to manual focus can
> dramatically reduce the latency. I can detect no latency at all in
> manual
> focus mode on my camera.
>

I find manual focusing takes longer than the latency period for auto
focusing. However, when you're focusing for special effect, there's
nothing better than fine tuning by hand.

Marilyn

At 09:04 PM 11/29/2004 -0500, Marilyn Matty wrote:
>I find manual focusing takes longer than the latency period for auto
>focusing.

Oh, a great deal longer. I didn't mean to imply anything else. (And my
Minolta DiMAGE Z1 has a particularly clumsy manual focus.)

Manual focusing is useful when you know ahead of time what distance your
focus will be, at least close enough that it won't need to change, but you
need minimum latency when you actually take the shot. Examples:

-- Children playing in a small spot some distance away, in bright light.

-- Sports (especially races) where competitors will be in a specific spot
for a very short time.

-- Shots of flowers and plants when there's a breeze or the subject moves a
lot.

-- Close-ups where you plan to get as close as the camera allows but the
subject is likely to move (such as insects, or flowers in a breeze or
shifting light).

These situations can often be handled simply by locking the focus in
advance. This however -- at least on all the cameras I've tried -- requires
relocking it for each shot, which can be inconvenient.

In such cases you can set the manual focus in advance and decrease the
latency to just your own reaction time. But it's a limited solution, not a
general one. My camera stays on auto-focus (AF) about 90% of the time,
despite the fact that manual focus capability was one of my required
features when I went shopping. And I think that improvements in its focus
algorithms would allow me to keep AF on even more.

Of course, manual focus is also useful when the AF algorithm simply fails.

I was responding to a comment on time and power usage required for AF,
which was why I didn't expound on this the first time around. But Marilyn
handed me a soap box.

Edward