Version 2: Why is a large sensor better

Forgive me for starting a new thread, but the first one is now so full with confused arguments that I don't expect anyone to have the patience to sort out the interesting arguments. Besides, I think I have understood a way to say what I mean that is more natural for photographers. Lets try it!:

Large sensors are better because it is cheaper to build lenses with large apertures for them. Large apertures gathers more light and thus gives you less noise in your pictures.

It is the aperture that determines how many photons a camera detects, NOT the sensor size.

PS. I am assuming that we hold FOV constant in comparisons, ie, when comparing noise we compare pictures of the same things.

All agree?

Other benefits with large sensors? (For one thing, they have a higher saturation level and thus a larger max dynamic range.)

woland21 said:

Forgive me for starting a new thread, but the first one is now so
full with confused arguments that I don't expect anyone to have the
patience to sort out the interesting arguments. Besides, I think I
have understood a way to say what I mean that is more natural for
photographers. Lets try it!:

Large sensors are better because it is cheaper to build lenses with
large apertures for them. Large apertures gathers more light and thus
gives you less noise in your pictures.

It is the aperture that determines how many photons a camera detects,
NOT the sensor size.

PS. I am assuming that we hold FOV constant in comparisons, ie, when
comparing noise we compare pictures of the same things.

All agree?

Click to expand...

Well, maybe it's a little more clear now, but for many "aperture" means the same as "f-stop", and you can't really say that that's wrong, since...

Relative aperture = f-stop
Absolute aperture = aperture diameter

So, the confusion will continue

Steen Bay said:

Well, maybe it's a little more clear now, but for many "aperture"
means the same as "f-stop", and you can't really say that that's
wrong, since...

Click to expand...

OK, I mean aperture as measured in millimeters.

Steen Bay said:

So, the confusion will continue

Click to expand...

For sure!

This means that common statements on the theme "I wish that camera had a larger sensor so it would have better high ISO performance" are quite misguided.

What that person is actually wishing for is the same camera but with a larger aperture. That will give better high ISO performance. The sensor size doesn't matter and can be entirely left as an optimization for the manufacturer.

So, next time you go shopping for a fixed-lens camera with good high ISO performance, look for a camera with a lens with a large aperture. Forget about sensor size, it doesn't matter.

"Oh no! This can't be true, everyone is talking about sensor size!" Oh, yes, it is true.

For a DSLR the situation is a little different because the set of available lenses is already set. But here we have another problem, most lenses were made for projecting on film. Today these lenses often project much of their light outside the sensor in the camera, with predictable results for low light noise. So for DSLRs it becomes more important to look for a sensor that has a large fill factor, ie, a sensor that actually fits the available lenses.

(There is however ANOTHER good reason to look for a large sensor, and that is high dynamic range.)

...totally obscure.

Forgive me, but you're making this whole thing more and more incomprehensible by insisting on developing your own theory / explanation for the advantages of large sensors, and essentially repeating it over and over, even while people (in the other thread) were pointing you at the simple facts and explanations.

Larger sensors = larger or more sparse photosites = less "overflow" or whatever it's called between photosites.

That'll do it for me and I'm prepared to say, non-expert as I am, that lenses and aperture are just a red herring. I might as well say "Big fat lenses = better white balance", there's probably an element of truth in that, it's just that certain other factors ar just a tad more significant.

Maybe I'm wrong. But I don't think I've seen anyone who seems to know what they're talking about agree with your angle.

I think you should give it a rest, read up on the subject, and come back with a more polished theory. If you must.

--

Arrowman said:

Maybe I'm wrong. But I don't think I've seen anyone who seems to
know what they're talking about agree with your angle.

Click to expand...

You should read the other thread more carefully. Several very well educated and experienced people have agreed with me. But those are not the people that write 50 posts telling me I got this all backwards.

Arrowman said:

I think you should give it a rest, read up on the subject, and come
back with a more polished theory. If you must.

Click to expand...

No one has caught me with a single error in my reasoning yet. So I don't really know what I should read up on. A lot of people have just said that they don't like my comparisons, for different reasons. And that they think I talk about this in an unusual way.

The undisputable truth is that the number of photons collected by any fixed-lens camera on the market can be calculated roughly by:

1. photons = (aperture surface area)x(solid angle of FOV)x(exposure time)

No one can dispute this formula and no one has tried. Note carefully that you do not need to know the sensor size of a camera to know how many photons it detects in low light conditions!

But then you translate this very simple and undisputable formula to photographic units, and it becomes less clear. And suddenly a lot of people seriously believe that the physics of the situation has really changed just because you changed units, and that sensor size now matters for low light performance.

woland21 said:

It is the aperture that determines how many photons a camera detects,
NOT the sensor size.

Click to expand...

The sensor size determines how many photons a sensor can detect per unit time and in total.

The user determines how much light the sensor gets when he picks a lens and aperture.

With a larger sensor, the user will be able to achieve higher signal to noise ratio than is possible with a smaller sensor.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

Ron Parr said:

The sensor size determines how many photons a sensor can detect per
unit time and in total.

Click to expand...

No, the sensor size only determines the maximum number of photons the sensor can detect. This doesn't matter much under low light conditions.

Member said:

The user determines how much light the sensor gets when he picks a
lens and aperture.

Click to expand...

Yes, and usually the camera comes with a fixed lens. For DSLRs the user picks a lens.

Member said:

With a larger sensor, the user will be able to achieve higher signal
to noise ratio than is possible with a smaller sensor.

Click to expand...

In good lighting, yes.

What I have been saying all the time. Doesn't hurt to be reiterated though.

"The undisputable truth is that the number of photons collected by any fixed-lens camera on the market can be calculated roughly by:

1. photons = (aperture surface area)x(solid angle of FOV)x(exposure time) "

This is NOT the number of photons collected. It is proportional to the number AVAILABLE for collection. They may or may not actually be collected.

The silicon sensor technology in use today is limited to collecting a maximum of less than about 1300 photons per square micrometer. A sensor must be large if it is to collect many photons.

"Large sensors are better because it is cheaper to build lenses with large apertures for them. ....."

I'd sure like to know of a source for these inexpensive large aperture lenses!

Has anyone yet pointed out that, first and foremost, a large sensor is better for the same reason a larger piece of film is better? The first step is still analog. The captured image, whether on film or on a sensor, is a secondary image projected by a lens. There is no perfect lens, and the imperfections that are inherent in any lens produce imperfections in that projected/captured image. When we view the captured image, we generally magnify it, so the larger the captured image was in the first place, the less we have to magnify to the imperfections produced by even the finest lens.

There are, of course, additional factors that are unique to digital processing. But I'd be surprised if any of them has as great an impact as the basic optics of image size.

woland21 said:

No, the sensor size only determines the maximum number of photons the
sensor can detect. This doesn't matter much under low light
conditions.> >

Click to expand...

I was following these threads with some interest, but now I need to know precisely why it doesn't 'matter much'? You see I would have thought that if fewer photons were getting through then it would have been even more important to keep the S/N ratio as low as possible by using a large sensor on which the receptors were more widely spaced.

Do they not refrigerate sensors on big telescopes in the hope of maximising the signal data?

woland21 said:

--

Click to expand...

John.
Please visit me at:
http://www.pbase.com/johnfr/backtothebridge
http://www.pbase.com/johnfr

Let's consider the following story:

Woland and Ron are having a debate about whether big buckets capture more water than small buckets. Ron say big buckets capture more water, but Woland says they do not. So, they decide to conduct a little test. They both go out in the rain, Woland with a bucket of a particular diameter and depth, and Ron with a bucket of 5X the cross sectional area and the same depth. Ron and Woland come inside. They measure the amount of water they've collected and it appears that Ron has collected 5X as much water.

Now, Woland asks to do the experiment again, but this time he takes a funnel with him which is equal in diameter to Ron's bucket. They do the experiment and they now collect the same amount of water.

A stubborn fellow, Ron asks to do the experiment again. This time, however, he brings a funnel that is 5X the cross sectional area of Woland's, scaling the funnel size with his bucket size. They come inside and the results of the original experiment are repeated: Ron has collected 5X as much water.

Woland now insists that Ron is misled people by using a larger funnel, so he adjusts his funnel size to compensate. Now the two go back and forth in a funnel arms race, getting larger and larger funnels with each trial.

Eventually, the funnel manufacturer tells Woland that he can no longer manufacture funnels to Woland's specifications because it just isn't possible to make a structurally sound funnel of the desired diameter that channels all of the water into a bucket so small. Woland now insists that the experiments continue and that he should be allowed to use theoretical funnels instead of of real ones. They continue.

Finally, all of the fun ends when Woland returns with an overflowing bucket. Ron's bucket is also full, but has 5X as much water as Woland's because it's 5X larger.

Ron and Woland sit down and try to figure out what they have learned from this fun exercise. Woland concludes, "It is the funnel that determines how much water a bucket collects, NOT the bucket size."

Ron concludes, "Bigger buckets have the ability to capture more water. The existence of funnels doesn't change the way buckets work - it just gives you a way of compensating, within a limited range, for the discrepancy between buckets."

Now, hopefully most of you have realized that rain is energy, buckets are pixels, and funnels are lenses.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

woland21 said:

Ron Parr said:

The sensor size determines how many photons a sensor can detect per
unit time and in total.

Click to expand...

No, the sensor size only determines the maximum number of photons the
sensor can detect. This doesn't matter much under low light
conditions.

Click to expand...

To be more precise:

At constant illumination the sensor size determines the the number of photons the sensor can detect per unit time.

The sensor size also sets an upper bound on the number of photons the sensor can detect at any illumination level.

woland21 said:

woland21 said:

With a larger sensor, the user will be able to achieve higher signal
to noise ratio than is possible with a smaller sensor.

Click to expand...

In good lighting, yes.

Click to expand...

The peculiar thing is that in your universe, this is true for ANY illumination level. Why? In your universe, we can always pick a wider aperture lens to compensate for the smaller sensor's difficulty capturing photons. However, if we can put wider aperture lenses on the small sensor in this universe, then we can also do it for the big one. Therefore, we can always assume that we have the ability to pick lenses that will push the small sensor to saturation, but keep the big sensor well within limits.

Ironically, this is all a consequence of your implicit assumption that you can always compensate for the weak light gathering ability of a small sensor by exposing it to brighter light.

Interestingly, in the real world, the big sensor fairs the best in poor light because real lenses have roughly the same min f/stop.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

Dave Martin said:

"Large sensors are better because it is cheaper to build lenses with
large apertures for them. ....."

I'd sure like to know of a source for these inexpensive large
aperture lenses!

Click to expand...

I didn't say that they would be cheap, I said that they would probably be cheaper than the same aperture lenses for small sensors. Isn't that right? I am no expert in lens construction.

Dave Martin said:

This is NOT the number of photons collected. It is proportional to
the number AVAILABLE for collection. They may or may not actually be
collected.

Click to expand...

Yes of course, this formula is only valid under low light conditions where you do not saturate the sensor.

Dave Martin said:

The silicon sensor technology in use today is limited to collecting a
maximum of less than about 1300 photons per square micrometer. A
sensor must be large if it is to collect many photons.

Click to expand...

Yes, this has been cleared out many times already, and I have agreed whole heartedly the whole time. Under good lighting you can get better dynamic range from a larger sensor. I am NOT saying that larger sensors are not better. Quite the contrary, that was where I started the discussion, by asking WHY larger sensors are better.

john farrar said:

woland21 said:

No, the sensor size only determines the maximum number of photons the
sensor can detect. This doesn't matter much under low light
conditions.> >

Click to expand...

I was following these threads with some interest, but now I need to
know precisely why it doesn't 'matter much'? You see I would have
thought that if fewer photons were getting through then it would have
been even more important to keep the S/N ratio as low as possible by
using a large sensor on which the receptors were more widely spaced.
Do they not refrigerate sensors on big telescopes in the hope of
maximising the signal data?

Click to expand...

Those sensors are refrigerated either because they are trying to detect IR light (and therefore want to minimise heat emitted by the camera) or are trying to reduce thermal and readout noise in the imaging chip. It doesn't affect the ability to detect signal but it does improve noise control.

Lyle Aldridge said:

Has anyone yet pointed out that, first and foremost, a large sensor
is better for the same reason a larger piece of film is better? The
first step is still analog. The captured image, whether on film or on
a sensor, is a secondary image projected by a lens. There is no
perfect lens, and the imperfections that are inherent in any lens
produce imperfections in that projected/captured image. When we view
the captured image, we generally magnify it, so the larger the
captured image was in the first place, the less we have to magnify to
the imperfections produced by even the finest lens.

Click to expand...

Wow, a constructive post! This is indeed another example of why a larger sensor is better, in addition to increased dynamic range under good lighting.

john farrar said:

woland21 said:

No, the sensor size only determines the maximum number of photons the
sensor can detect. This doesn't matter much under low light
conditions.> >

Click to expand...

I was following these threads with some interest, but now I need to
know precisely why it doesn't 'matter much'? You see I would have
thought that if fewer photons were getting through then it would have
been even more important to keep the S/N ratio as low as possible by
using a large sensor on which the receptors were more widely spaced.
Do they not refrigerate sensors on big telescopes in the hope of
maximising the signal data?

Click to expand...

That is right. Surprising thing is that high res sensors, like FZ50 still seem to manage the smaller pixels very well. I have posted a link that proves this several times.

Sooner or later, if pixel density goes up, there will come a point when this becomes important. But we don't seem to be there yet.

A bucket 5 x diameter will collect about 20x more rain, not 5x as much.

Pi x radius squared.

This and the rule of squares is ofted overlooked on these forums.

There also is a need to distinguish between sensor and sensor array.
--

'While not exactly disgruntled, he was far from feeling gruntled. He spoke with a certain

what-is-it in his voice, and I could see that, if not actually disgruntled, he was far from
being gruntled.'
~ Pelham Grenville Wodehouse (1881-1975)

Ron Parr said:

Let's consider the following story:

Woland and Ron are having a debate about whether big buckets capture
more water than small buckets. ...

Click to expand...

A good analogy, but you got it wrong. This is the way it should read.

Ron and Woland has one funnel each. Both the funnels are the same diameter. Both funnels are connected by a hose to a bucket. However, Ron's bucket is larger than Wolands.

Ron says my bigger bucket collects more water!

Woland says yes, but only if my bucket overflows. If it is only a light rain my smaller bucket will collect the same amount of water, since the diameter of our funnels are the same.

The go out and try, and guess who is right?