SLR v P&S

AlanL wrote:
In article ,
David J Taylor
wrote:
Alan,

You say "if the aperture in mm is the same".

If consider the 4/3 system as being half the linear size of a
full-frame sensor, then for a give image diagonal angle, the
required physical focal length is half that of the 35mm system, and
if the lens aperture is constant, then the f/number will be half
that of the 35mm lens. So a 4/3 system with an f/2 25mm lens has
the same sensitivity as a full-frame system with an f/4 50mm lens.

So if you want the same sensitivity as a 50mm f/1.8 lens, you would
need a 25mm f/0.9 lens.

Is this what you are saying?

Exactly. In both cases, the lens objective diameter
would be the same, would snag the same number of photons,
and the sensitivity would be the same, regardless of
the sensor size -- except that the dSLR sensor needs to
be full frame, not smaller. If it is smaller, it will
have less sensitivity.

--alan

OK, Alan, but doesn't raise the question: "Where do I get my f/0.9 lens to
get the same sensitivity?"

David

In article ,
David J Taylor wrote:

Exactly. In both cases, the lens objective diameter
would be the same, would snag the same number of photons,
and the sensitivity would be the same, regardless of
the sensor size -- except that the dSLR sensor needs to
be full frame, not smaller. If it is smaller, it will
have less sensitivity.

--alan

OK, Alan, but doesn't raise the question: "Where do I get my f/0.9 lens to
get the same sensitivity?"

Sure, of coarse. But you can't get it if everyone
thinks that sensitivity is determined by sensor size.
The manufacturers make what people buy, and if people
think that sensor size is the end-all-be-all, that's
what manufacturers are going to spend their efforts on
rather than the appropriate light-gathering optics.
Right?

-AlanL

In article , AlanL
wrote:

Sure, of coarse. But you can't get it if everyone
thinks that sensitivity is determined by sensor size.
The manufacturers make what people buy, and if people
think that sensor size is the end-all-be-all, that's
what manufacturers are going to spend their efforts on
rather than the appropriate light-gathering optics.

the lens has nothing to do with it. take a crop sensor slr, put a dx
lens set to f/8 on it then swap it with an fx lens set to f/8. there
won't be any difference in noise in the image.

AlanL wrote:
In article ,
dj_nme wrote:
AlanL wrote:
In article ,
MC wrote:
It doesn't really matter what the reason is for using a certain size of
sensor, squeezing the same amount of pixels onto a small sensor in favour of
a big one will (at present technology levels) will cause slightly inferior
image quality from the smaller sensor.
Why does smaller pixel make a inferior image
and how much is 'slightly' ?

--alan
Smaller pixel size means less active area to detect the image and
therefore greater amplification required to achieve the same
sensitivity, both causing greater noise (or "digital grain") to show up
in photos.
That's what causes the red/green/blue speckles in shadow areas.

I think you didn't read my first post. Larger area
is good for dSLRs using 35mm lenses that were designed
to focus an image over a 43.27mm disk at the focal plane.
If the sensor is smaller than full frame, a lot of the
photons will hit outside the sensor.

That's not relevant to noise appearing in photos.
What it can cause is a lowering of image contrast if the mirror box of
the crop-frame DSLR camera isn't properly flocked (coated in a black,
non-reflective surface).

If, however, the lens is designed to match the sensor
size, each photosite will receive a greater number of
photons. Usually the optics match the sensor size for
P&S or micro-four-thirds cameras, but with many dSLRs
the optics are oversized and photons are wasted. I.e.,
for dSLRs larger sensors mean less photon waste, and
better sensitivity.

Smaller pixel = less sensitive area = less light gathering per pixel =
more noise compared to signal.

So my question is, assuming the optics match the sensor,
why would a smaller sensor give an inferior image, and
by how much? That was the claim.

See above.

AlanL wrote:
David J Taylor wrote:
AlanL wrote:
MC wrote:
It doesn't really matter what the reason is for using a certain size
of sensor, squeezing the same amount of pixels onto a small sensor
in favour of a big one will (at present technology levels) will
cause slightly inferior image quality from the smaller sensor.
Why does smaller pixel make a inferior image
and how much is 'slightly' ?

Read the articles on Roger Clark's Web site:
http://www.clarkvision.com/new_articles.html
such as:
http://www.clarkvision.com/imagedeta...l.size.matter/
http://www.clarkvision.com/imagedeta....size.matter2/
http://www.clarkvision.com/imagedeta...mance.summary/

That backs my arguement up. It's saying if the sensor is
smaller than the optics were designed for, it's going to
perform worse. Well yes, of course, because photons are
being focused outside the sensor, and bigger sensors will
catch more of the photons. With many dSLRs, photons
are just being thrown away because the optics do not match
the sensor.

But suppose you are talking about a micro four-thirds camera.
The optics are designed to match the sensor size so the photon
waste will be minimal. The photosite area will be smaller,
but the photon flux will be higher resulting in the same
number of photons being caught as in a full frame dSLR with
35mm lenses -- assuming the same objective size or aperture
size (measured in mm, not f/#). Sensitivity, I maintain,
is a function of how many photons are brought into the system
(objective size) and the efficiency of their use (minimizing
absorption losses, matching optics design to the sensor size,
etc.), but not directly the sensor size. The sensor size
thing is just an artifact of wanting to use optics designed
for 43.27mm focal disks on dSLRs, which is irrelevant for non-
dSLRs or dSLRs for which the optics match the sensor size.

I don't know Alan, that's reasonable logic though everything I've heard
says it doesn't work. What you are saying is if the lens is designed to
funnel that light into a smaller image circle, then you've got the same
amount of light packed onto a smaller sensor & nothing is cropped off.

Conceivably the micro-4/3 mount with it's shorter flange to sensor
distance takes advantage of that but we would have heard that claim if
it were possible. There is some benefit but not much. I *have* heard
many times that for longer focal lengths, you can't save much bulk at
all going with a cropped image circle design so I guess there's no way
to capture that lost edge info. Perhaps the answer is something like, if
the lens is designed to funnel that light into a smaller image circle,
then the effective focal length has been changed, like adding a
teleconverter - which decreases the amount of light and f/stop by cropping.

I can't say why but this sounds like a perpetual motion machine proposal
trying to get more light into a smaller sensor. Maybe start a new thread
if you want to pursue it further & nothing better comes up here. It
would be nice to have an understandable answer.

--
Paul Furman
www.edgehill.net
www.baynatives.com

all google groups messages filtered due to spam

bugbear wrote:

I'm not sure what 1:1 macro means in the era of digital;
"sensor pixels per subject mm" seems a more useful measure.

Yeah, I don't even know what ratio we are talking about. At one point I
heard mention of a 40mm subject capture filling the frame but a couple
mm away, then mention of easily shooting a 35mm slide... If you shoot a
mm scale, how many mm fit across the frame and how many pixels? Say you
print at 300dpi and figure a ratio from that to even out the pixel
count. Then the question is what kind of detail can you capture at that
magnification? Measure that the way lens tests are done, with MTF, in
lppmm line pairs per millimeter but measure it on the print. That's your
comparable number. I suspect the results will respond to the laws of
physics and the larger DSLR will win.

Doug Jewell wrote:
... But there is a catch. The Ixus 70 achieves that magnification when the subject is a mere 3cm from the lens.

This is always true for extreme macro

I did a test the other day with a 100mm macro on about 200mm of bellows
extension and got about... well, I'm not even sure what magnification
but way over 1:1 and had about 2 to 4 inches of working distance.

Here's with a tiny webcam attached (most of the shots):
http://edgehill.net/Misc/photography...bcam-macro/pg1
Here's with a full frame DSLR (cropped to the same pixel count):
http://edgehill.net/Misc/photography...m-macro/pg2pc9

This happened to be a back-lit LCD screen but I shot a metal ruler also
in dim light with long exposures... there isn't any big difference
except more noise on the webcam. Reduce the webcam by 50% & they are
about the same size & same detail captured with this easy even LCD back
lighting. I don't know grin. I guess my point is you can have ample
working distance but you need a long lens and you'd never see a lens
that size on a P&S or that much pixel density on a DSLR grin. The test
was about working distance.

The pixels on the screen photographed are at 90dpi, 3 colors wide in an
RGB pixel. I could have used the 105mm lens to get twice the
magnification on the DSLR shot, it's not really scientific, just playing.

--
Paul Furman
www.edgehill.net
www.baynatives.com

all google groups messages filtered due to spam

AlanL wrote:
In article ,
David J Taylor
wrote:

Exactly. In both cases, the lens objective diameter
would be the same, would snag the same number of photons,
and the sensitivity would be the same, regardless of
the sensor size -- except that the dSLR sensor needs to
be full frame, not smaller. If it is smaller, it will
have less sensitivity.

--alan

OK, Alan, but doesn't raise the question: "Where do I get my f/0.9
lens to get the same sensitivity?"

Sure, of coarse. But you can't get it if everyone
thinks that sensitivity is determined by sensor size.
The manufacturers make what people buy, and if people
think that sensor size is the end-all-be-all, that's
what manufacturers are going to spend their efforts on
rather than the appropriate light-gathering optics.
Right?

-AlanL

Not the way I see it, Alan. You can't get an f/0.9 lens off-the-shelf
because it's too expensive to manufacture, and if you make the sensor
smaller again (such as is in a P&S camera) you require optics which simply
cannot be made. So the smaller sensor leads to lower sensitivity i.e.
poorer signal-to-noise ratio at the same light level.

Cheers,
David

"Paul Furman" wrote in message
...
bugbear wrote:

I'm not sure what 1:1 macro means in the era of digital;
"sensor pixels per subject mm" seems a more useful measure.

Yeah, I don't even know what ratio we are talking about. At one point I
heard mention of a 40mm subject capture filling the frame but a couple mm
away, then mention of easily shooting a 35mm slide...

In the olde days of film I remember Macro being defined as 1:1 or greater
anything less
was close focus and 1:1 meant an object (insect etc) measuring say 20mm
would produce an image measuring 20mm on a negative whether the neg. was
110,
35mm, 2 1/4 square or half plate or whatever.

David J Taylor wrote:
Neil Harrington wrote:
[]
It doesn't, necessarily, except in low light. Then smaller pixels
collect fewer photons and therefore the image has to be amplified
more. Greater amplification produces more "noise" which tends to
destroy image quality.

It's not so much a question of amplification, although that does come
into it, but that a small number of photons (such as the number
collected by a smaller sensor) will inherently have a higher noise
level, and hence a poorer signal-to-noise ratio. I.e. the noise is
in the photon stream itself.

http://en.wikipedia.org/wiki/Shot_noise

Cheers,
David

That's most interesting. Point taken.

Neil

whisky-dave wrote:
"Paul Furman" wrote in message
...
bugbear wrote:

I'm not sure what 1:1 macro means in the era of digital;
"sensor pixels per subject mm" seems a more useful measure.

Yeah, I don't even know what ratio we are talking about. At one
point I heard mention of a 40mm subject capture filling the frame
but a couple mm away, then mention of easily shooting a 35mm
slide...

In the olde days of film I remember Macro being defined as 1:1 or
greater anything less
was close focus and 1:1 meant an object (insect etc) measuring say
20mm would produce an image measuring 20mm on a negative whether the
neg. was 110,
35mm, 2 1/4 square or half plate or whatever.

I guess it gets down to "what are you trying to communicate when you
say "macro"?

Naively I used to think that pixels per subject millimeter (didn't
think of it with quite that precision but it's a useful measure--thank
you) would be the defining metric. But as I've learned more I've come
to realize that there are other effects attendand upon macro shooting
that may make that metric less useful than the dimensional 1 mm of
subject is 1 mm of sensor. One in particular is depth of field. On a
typical point and shoot it is not possible to isolate a portrait from
the background with depth of field--optically you're shooting with the
subject near infinity focus and well inside the hyperfocal range due
to the large disparity between subject size and sensor size. With an
APS-C or "full frame" sensor and a fast lens you can do this if there
is reasonable distance between subject and background. Go to an 8x10
though and a child's is about the same size as the frame--effectively
portraits beome macro or near-macro shots, with an inch of face being
an inch of film, and now you're working so close, optically speaking,
that depth of field is virtually nonexistent, to the point that parts
of the face can be isolated from other parts using depth of field.

So that's one way of looking at it. Another though is that when most
of us think of "macro" the image that pops into our heads is an insect
or flower or leaf or coin or some other small subject. With an 8x10,
filling the frame with such a subject would be way beyond macro, with
a 35mm it would be around 1:1 or a bit beyond, with a point-and-shoot
coins would not be even up to 1:1 although an insect might.

So the question becomes which definition is most useful, the one in
which a mm of subject is a mm of on-sensor image or the one in which
the subjects are small?

--
--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)