"Ed Anson" <EdAnson@comcast.net> wrote in message news:3F18A45E.7070407@comcast.net...

> David Ruether wrote:

> > "Ed Anson" <EdAnson@comcast.net> wrote in message news:3F175D3F.5080604@comcast.net...

> >>In a practical system, all else cannot be equal.

> > But in an experimental system, to determine the effect of changing

> > *one* aspect only (as here, with a question only of the effect of

> > changing the sensor size [only...;-]), you *can* make all else equal.

> Not *all* else. No way.

[This is getting silly - you need to know more about optics, and

experimental procedures, but, here goes...;-]

Yes, you can.

You can take the raw CCDs, power them, and measure their

outputs when exposed to the same light. Again, the question is

about the relative sensitivity of CCDs of different size (alone),

with all else (reasonably!) specified as equal. This *should* be

an easy concept. And, the lens *especially* is irrelevant since it

is so simple to make the two lenses both cover each CCD and

deliver equal light intensity to the CCDs...

> For example, if you make the optical systems the same, then the image

> focused on the two sensors is different. You can't avoid that, not in

> this universe. You can make the images the same (except for size) only

> by making the optics different.

Different, but entirely equal for the purposes of this discussion!

As I pointed out before (differently...), you can have a 300mm f5.6

lens cover 8"x10" film, with the same angle of view as a 45mm f2.8

lens that covers only 24mm x 36mm film area - and BOTH can be

set to f8, for instance, and BOTH will deliver to the different film

areas the same angle of view and light intensity (aimed at the same

subjects, under the same illumination). Most zooms made for video

are about the same speed, can be set to the same relative aperture,

and can cover the same angles of view. Since two lenses are

effectively equal for two different CCD sizes, their differences in

effect (and their need to be considered) disappear, by definition.

GONE...! ;-)

> In an experiment (or even a thought experiment) you hope to hold the

> most relevant variables constant. But that implies consideration of what

> is relevant. Reasonable people may differ on that. So any discussion of

> comparisons must make clear what is considered relevant.

But those things that drop out due to equality of effect should

be discarded...

> When comparing sensors, you can hold the optics constant or you can hold

> the image and total collected light constant. You can't do both, and

> either choice gives a different result.

This is not correct... And I don't understand why you cannot see this...

> IMHO the latter choice is more relevant to the OP's question.

It would be equally relevant, if it were relevant (and that was what

was asked about), but it is not, and it was not...;-).

> > If after doing this, you want to add extra questions, that is OK, but

> > irrelevant to the basic single question that can be answered easily

> > by stripping away what is not useful in answering the basic question.

> > Or, if I want to include the irrelevant lens, associated electronics, etc.,

> > why not also argue about the color of the camera grip, whether or

> > not it is raining during the experiment, etc., etc., etc....;-)

> I suppose you could try to keep the grip, weather and phase of moon

> constant, but I suspect those won't affect the result very much. ;-)

Yes, along with trying to make the lens relevant, when it may be only

in exceptional cases - but it would be hard to find these cases, and

hard to reasonably define them in a way that affects the result...;-)

> OTOH keeping the image constant (or not) affects the result, so it's

> important to be clear about whether you are doing that and whether your

> choice enables you to answer the question at hand.

??????????????

> >>As I pointed out before, if identical optics are used then the smaller

> >>sensor records a smaller portion of the image than the larger one, so

> >>the framed image is not equal. In this hypothetical situation, the

> >>smaller sensor captures less light per pixel. In that sense, it is less

> >>sensitive.

> > This is incorrect as an arguement,

> Pardon me for interrupting, but I was agreeing with you here. I was

> simply being explicit about what was constant (and what was not) and

> arriving at the conclusion you are arguing for. So how is that an

> incorrect argument?

Because it is an incorrect view of the situation in that it doesn't

matter what part of the image of the lens is used (or not) - it is the

area illuminated (equally for the two) by *any* means, including

(especially, to simplify the discussion) a uniform diffuse light source

that is equal in intensity for both (a redundancy, but...;-).

> > since you can easily specify

> > that the lens will cover both sensor sizes, that it is a zoom of

> > constant relative aperture, and that it can be zoomed to cover

> > the same angle of view with both sensors.

> Here you propose changing the optics, making them not equal. Put another

> way, you are opting to change both the focal length and the iris

> diameter to keep the image intensity constant while changing the image

> size and the total amount of light collected. This would give the result

> you want, but you can't say you're keeping *all else* constant when you

> are changing those things.

Yes you can, in that the effects of the lenses for this experiment are made

equal. I'm not saying that the lenses physically are the same (though they

could be), but that most lenses likely to be used *would* be equal in

their relevant effects on the two CCDs, set in a reasonable way (mainly,

to the same relative stop). They could then *easily* both deliver the same

light intensity to both CCDs, given the same light sources and angle and

width of view of them...

> Of course, if you only consider image intensity to be relevant, that is

> your choice. But I don't think that choice is consistent with the OP's

> question.

But it is basic to seeing what effect area has. Without specifying that the

light intensity is equal, nothing else makes sense in this discussion.

And, this is so easily done "real world", too...;-)

> > The lens drops out

> > as an issue. If you specify different lenses that cannot be used

> > equally, you have made the question of sensor size effect not

> > directly answerable by introducing an irrelevant "monkey wrench",

> > but that is silly... If you want, you can assume that almost all

> > zoom lenses for video can be set to cover the equivalent angle

> > of view of, say, a 70mm lens in 35mm, and that all zoom lenses

> > can be set to a true f5.6 relative aperture. In this (most common)

> > case, the lenses perform equally on the two sensor sizes, and

> > they do drop out as issues...

> But the lens was an explicit part of the OP's question.

Then please quote it, if you have it - but, again, one can say that

an f1.2 lens will pass a greater light intensity than an f3.5 lens, but

that tells me nothing about the relative sensitivities of two different

sized CCDs (all else equal...;-). And, as it happens, most zoom lenses

for video cameras are about the same "speed" (there are exceptions,

but all can easily me made about equal, simply by selecting a medium

stop for all...).

> >>But in real life the optics would be adjusted to frame the same image,

> >>and that complicates things. To make the image equal, we must make the

> >>optics different. So "all else being equal" isn't really possible.

> > Sorry, but this is nonsense...

> Such comments are not productive.

But it is true - this made no sense ("nonsense"), as I have pointed out

repeatedly, and again, above. We are specifying that all "conditions" are

equal except CCD size, not that *everything* is equal - this would make no

sense, since we cannot jam (or power, or receive and process the signal of)

a larger CCD into a small-CCD camera, etc. We need only specify that

all conditions are equal but the exception under question, if this is reasonably

done - and in this case, it is...

> >>>Since this is a discussion of CCD size

> >>>alone vs. sensitivity, only the CCD area need be a concern.

> Perhaps we are in a different discussion. The OP's question had to do

> with differences in low light performance, given reasonable adaptations

> of the optics. Sensitivity of the sensor is relevant to that question,

> but not the only factor.

Yes, of course. The CCD type also has considerable effect, as does the

processing of the signal. But most lenses are about the same "speed",

and can be set to the same stop, leaving other questions open...

> >>> RGB's checkerboard in the rain covers the issue well enough,

> It is an excellent description of why a smaller sensor produces a

> smaller signal, given equal light intensity. If that were *the issue* I

> wouldn't have jumped in here.

Then we are talking about different things, but the title of this

thread does happen to be, "Why would a larger sensor give

better low light performance?" This is a rather clear and

"clean" question...;-)

> The OP was not positing equal light intensity. He seems to be thinking

> of optical systems that focus the same amount of light on each sensor,

> so that (presumably) each pixel receives the same amount of light on

> both sensors, albeit at different intensities. The checkerboard analogy

> fails to explain why the larger sensor still performs better in this

> situation. [I have attempted to explain that in a separate posting, in

> direct response to the OP.]

> <NOTE: The remainder of this dialog is clipped, because it seems unduly

> repetitive to me. I think I've made my point.>

And I, mine...;-)

--

David Ruether

d_ruether@hotmail.com

http://www.David-Ruether-Photography.com

Hey, take a gander at www.visitithaca.com, too...!