Why is a large sensor better?

[Ron Parr]

The problem is see with your notion of a "proper" comparison is that
it's a one way comparison due the impossibility of matching the large
sensor performance on the small sensor.

But a large sensor does match the performance of a small sensor for
equivalent images.

While the case where the DOF is equal is interesting, I don't agree with the notion that it is the only proper comparison. I think things are more interesting in the case where the f/stops are equal. These are the cases we actually want to be comparing to understand the differences, so I'm quite reluctant to say that all the most interesting comparisons improper.

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

 

[joe mama]

While the case where the DOF is equal is interesting,

It's interesting that you say "interesting". As I said earlier, it's curious to me that DOF plays less a role for some people's photography than does noise. I mean, I shoot wide open all the time, but that's because I love shallow DOF, not because I'm trying to reduce noise -- that's just a bonus. I shoot wide open even in bright light, having to drop down to ISO 50 just to be able to not overexpose the image.

Critics of FF always say how the DOF of FF is too shallow, failing to realize that you only need to stop down to achieve the same DOF with the same noise. ISO 100 on FF is not the same as ISO 100 on 1.6x -- ISO 250 on FF is the same as ISO 100 on 1.6x (for sensors of the same design and generation).

Because people compare images at the same f-ratios, rather than the same apertures (DOF), many myths of FF constantly get repeated.

I don't agree with the notion that it is the only proper comparison.

Oh, it's not the only proper comparison. I can see that you're resisting reading the essay I linked, since I've devoted no less than a whole chapter to exactly that:

http://www.josephjamesphotography.com/equivalence/

"SITUATIONS WHERE VIOLATION OF EQUIVALENCE IS VALID:"

I think things are more interesting in the case where the f/stops are equal.

It makes for a haphazard evaluation of systems. It's like someone else saying "I think noise adds to the photo, therefore noisier cameras are better."

These are the cases we actually want to be comparing to understand
the differences, so I'm quite reluctant to say that all the most
interesting comparisons improper.

Be that as it may, it is still "improper" to take a pic of a scene with a 5D at 50mm, f / 2.8, ISO 100 and talk about how it has softer edges and more vignetting, but less noise than an image from a 30D at 30mm, f / 2.8, ISO 100, when you could simply stop the 5D down to f / 4.5, ISO 250, and get an image with the same vignetting, same (or better) edges, and same noise.

Make the "proper" comparison (equivalent), and then discuss the options that are available for the larger sensor camera that are not available to the camera with the smaller sensor.

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

Please feel free to criticize, make suggestions, and edit my photos. If you wish to use any of my photos for any purpose other than editing in these forums, please ask.

 

[Jerry R]

Joe could you please elaborate on the comment about ISO in your previous post.

"ISO 250 on FF is the same as ISO 100 on 1.6x"

 

[Ron Parr]

I think I see at least one source of confusion. When you say "size of lens" I would guess that you actually mean "(physical) size of aperture".

Your version is still logically inconsistent in places, but if we stick with the version where you are playing the same game Joe is, i.e., insisting upon keeping the aperture small even when you scale up to a longer focal length lens, you are still constructing an artificial situation where you intentionally starve the large sensor for light.

To pick a nice round number, let's assume you're comparing a P&S with a digital SLR, so the larger sensors is about 5X as big. Just for fun, let's assume that the P&S is shot at f/8 (yes, I realize that's on the small side but it is possible for many cameras) and a focal length of 10mm. On the SLR, you'll need to use a 50mm lens and shoot at f/40. You'll need to do quite a bit of searching to find such a lens, if it's even possible since f/40 is quite small for a 50mm lens. Let's assume you do find it. Now you'll want to shoot at the same shutter speed to do a fair comparison, but to do this, you'll need to boost the ISO on the digital SLR dramatically, only using a fraction of the digital SLR's sensor capacity and testing the quality of the camera's amplifier to see if it can compensate.

So, now you've jumped through all of these hoops to starve your big sensor for light (by spreading the light that hits a smaller sensor over a wider area) and you conclude from this allegedly simple, fair, and natural comparison that big sensors don't capture more light.

In contrast, you have described the experiment where both sensors receive light at the same intensity, i.e., the case where the images are shot at the same ISO, same f/stop, same shutter speed, etc. as "misleading".

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

 

[joe mama]

Joe could you please elaborate on the comment about ISO in your
previous post.

"ISO 250 on FF is the same as ISO 100 on 1.6x"

Best to read the whole essay, of course, on "Equivalence":

http://www.josephjamesphotography.com/equivalence/

The chapter entitled, "EXPOSURE, TOTAL LIGHT, AND NOISE" explains it in detail. I'll give a quick synopsis/example here:

An image created on a 30D at f / 2.8, 1/100, ISO 100 will be created with the same total light as an image created on a 5D at f / 4.5, 1/100, ISO 250. When the two images are then resampled to the same output size , they will be equivalent (there are complications in the "resampling" that are mentioned in the chapter), and, if the sensors have the same design and generation, they will have the same noise.

The following article, linked at the bottom of my essay:

http://www.clarkvision.com/imagedetail/does.pixel.size.matter/

has a much more in-depth treatment of noise. However, it doesn't deal with different sensor sizes directly (but it does do so indirectly), so my next essay will be on noise to explain in even more detail.

Be sure to also take a look at the images that provide evidence for the assertion linked at the bottom of the essay. They show a *istunning* correlation with theory and practice.

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

Please feel free to criticize, make suggestions, and edit my photos. If you wish to use any of my photos for any purpose other than editing in these forums, please ask.

 

[Ron Parr]

Critics of FF always say how the DOF of FF is too shallow

I've never seen a reasonable person make such a claim. Then again, I do realize this is a no true Scotsman argument. ;-)

I think things are more interesting in the case where the f/stops are equal.

It makes for a haphazard evaluation of systems. It's like someone
else saying "I think noise adds to the photo, therefore noisier
cameras are better."

I don't see what's haphazard about a comparison designed to elucidate key differences.

These are the cases we actually want to be comparing to understand
the differences, so I'm quite reluctant to say that all the most
interesting comparisons improper.

Be that as it may, it is still "improper" to take a pic of a scene
with a 5D at 50mm, f / 2.8, ISO 100 and talk about how it has softer
edges and more vignetting, but less noise than an image from a 30D at
30mm, f / 2.8, ISO 100, when you could simply stop the 5D down to f /
4.5, ISO 250, and get an image with the same vignetting, same (or
better) edges, and same noise.

I don't seen anything "improper" about that. The only that seems improper to me is failure to point out that there's no way for the 30D to achieve the same effect as the 5D - unless Canon has a really fast 30mm lens that I don't know about.

Make the "proper" comparison (equivalent), and then discuss the
options that are available for the larger sensor camera that are not
available to the camera with the smaller sensor.

My opinion is that proper comparisons are designed to elucidate differences, not hide them. My impression is that your notion of a proper comparison involves working to hide differences by starving the big sensor then boosting the ISO to compensate.

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

 

[joe mama]

Critics of FF always say how the DOF of FF is too shallow

I've never seen a reasonable person make such a claim. Then again, I
do realize this is a no true Scotsman argument. ;-)

Oh man, you should get out more often! : ) Fallacious claims are rife in these fora. Here, let me link you to one I have handy:

http://forums.dpreview.com/forums/read.asp?forum=1029&message=24412696

That guy is pretty vocal, too. So is Rriley in the Olympus SLR Forum.

My opinion is that proper comparisons are designed to elucidate
differences, not hide them. My impression is that your notion of a
proper comparison involves working to hide differences by starving
the big sensor then boosting the ISO to compensate.

See, from my way of looking at it, saying that a 5D + 50 / 1.4 is equivalent to a 30D + 30 / 0.9 does highlight the differences from the systems. And, for the people who say that they have no need for f / 0.9 at 30mm (or any other f-ratio, for that matter), and thus 1.6x is better that FF 'cause it has sharper corners and less vignetting (false claims), then we can compare a 30D at 30mm, f / 2.8, 1/100, ISO 100 to a 5D at 50mm, f / 4.5, 1/100, ISO 250 and show that FF is, once again, at least as good (in terms of IQ).

In fact, whatever settings we apply to the 1.6x camera, we use the equivalent settings on the FF camera, and show, time and time again, that the FF camera is at least as good.

Equivalence does highlight the differences in a manner that is fair to both systems. And, yes, as we would expect, the larger sensor always does at least as well.

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

Please feel free to criticize, make suggestions, and edit my photos. If you wish to use any of my photos for any purpose other than editing in these forums, please ask.

 

[Ron Parr]

Joe could you please elaborate on the comment about ISO in your
previous post.

"ISO 250 on FF is the same as ISO 100 on 1.6x"

Here's a simpler way to think about it without the distraction of resampling.

An FF sensor has approximately 2.5X the area as a 1.6X sensor (1.6 x 1.6 = 2.56 ~ 2.5). So, a properly exposed image on an FF sensor will get 2.5X as much light as a properly exposed image on a 1.6X sensor. If you want to make an FF sensor perform like a 1.6X sensor, just starve it a bit by giving it 1/2.5th the light that would normally be required to get proper exposure.

What happens when you shoot at ISO 250 instead of ISO 100? You give the your FF sensor, 1/2.5th the light, all else being equal.

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

 

[Ron Parr]

Critics of FF always say how the DOF of FF is too shallow

I've never seen a reasonable person make such a claim. Then again, I
do realize this is a no true Scotsman argument. ;-)

Oh man, you should get out more often! : ) Fallacious claims are
rife in these fora. Here, let me link you to one I have handy:

http://forums.dpreview.com/forums/read.asp?forum=1029&message=24412696

That guy is pretty vocal, too. So is Rriley in the Olympus SLR Forum.

I didn't say that I never saw anybody make such a claim. I said that I never saw a reasonable person make such a claim.

Just for kicks.

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

I won't insist that you read it, or browbeat you if you don't. ;-)

My opinion is that proper comparisons are designed to elucidate
differences, not hide them. My impression is that your notion of a
proper comparison involves working to hide differences by starving
the big sensor then boosting the ISO to compensate.

See, from my way of looking at it, saying that a 5D + 50 / 1.4 is
equivalent to a 30D + 30 / 0.9 does highlight the differences
from the systems.

It's an interesting rhetorical device to appeal to imaginary lenses and non-existent ISO settings to make your point , but I think it's more compelling to focus on what can be shown in real photographs rather than imaginary ones. (You need non-existent ISO settings if you ask what the "equivalent" on the 30D is for an ISO 100 5D shot.)

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

 

[John Sheehy]

You are wrong.
For a f-stop the illumination of the sensor it's the same so both
sensor will get the same amount of light falling on them. But if both
have the same pixel count an individual photosite will be much larger
in a larger sensor (comparing the crop factor 1.6 and 6 sensors the
photosite is 14 times larger in the 1.6 crop factor sensor). So the
larger sensor has 14 more light falling on a photosite and thus a
higher SNR so less likely to fall below the threshold of noise being
observable).

That's true for the individual pixels, as measuring devices, but for the image, big pixels don't do much of anything for noise, but lose resolution.

--
John

 

[vor]

You wrote:

"The only assumption I make is that you hold the lens size

fixed, while changing the sensor-size." and "Keep the size of the
lens constant. Keep the same FOV, and then change the sensor-size."

If you mean that a FF-sensor with a 100mm, f/4 lens collects the same
total amount of light as a 2x crop sensor with a 50mm, f/2 lens, then
that's correct, but when you say "lens size fixed" and "lens
constant", then it doesn't make much sense!

It's obvious (at least to me) that by "lens size" he means aperture. A 100mm f/4 lens and a 50mm f/2 lens both have 25mm apertures. In a simple lens, the aperture dictates the diameter of the lens, whereas the focal length dictates the curvature of the lens. And for f/1.0 and slower lenses, the diameter of the lens dominates the size of the lens, by far.

Victor

 

[Jerry R]

OK, glad I did not criticize before I fully understood what you were saying. Large sensor cameras gather more light than small sensor cameras.

 

[David Banner]

All this would have been obvious to everyone if you could actually
buy any small lens/large sensor cameras. Although easy to construct,
these cameras don't make commercial sense. They would have to use
long focal lengths and long lenses and would probably provide the
buyer with very low price/performance.

And why don't we have any large lens/small sensor cameras? The large
lense is expensive enough that it doesn't make sense to put in a
small sensor. Also, such a lens would have to have very short focal
lengths and would again be very expensive.

Thus all cameras come with a lens that matches the sensor in size.
And because larger sensors always comes with larger lenses,
photographers have falsly come to belive that it is the larger sensor
that collects more light, when it is in fact the larger lens.

This is a very good explanation and it answers a question I've had for a while: why don't compact cameras use an APS-C sized sensor. The reason is that it'd need an APS-C sized lens.

 

[vor]

However, a larger sensor does not collect more light if you use it to
take a picture of the same thing, with the same size lens.

True, if the length of time of each exposure is the same.

The number of photons collected by a camera (and thus the number of
pixels that hits the sensor) is proportional to:

Pixels?

(the solid angle of the FOV) x (the area of the aperture)

I think you're missing shutter speed, or a statement that the above is per unit time. Shutter speed has to be the same between captures.

Thus, the size of the sensor has nothing to do with how many photons
that are collected. And, when the sensor sits behind a lens in a
camera you are wrong, a larger sensor does not collect more light.
The lens is perfectly capable of projecting all the light on just
about any size sensor.

But the size of the sensor combined with the focal length together dictate the FOV, and so you can derive one from the other. If you move focal length over to the aperture part instead of the FOV part (and focal length combines with aperture area to produce the intensity of light hitting the sensor), then you are essentially saying that intensity times sensor area determine the number of photons. Again, I would add shutter speed; combining that with the intensity (to produce exposure), you'd get that exposure times sensor area determine the total number of photons gathered.

Victor

 

[joe mama]

See, from my way of looking at it, saying that a 5D + 50 / 1.4 is
equivalent to a 30D + 30 / 0.9 does highlight the differences
from the systems.

It's an interesting rhetorical device to appeal to imaginary lenses
and non-existent ISO settings to make your point , but I think it's
more compelling to focus on what can be shown in real photographs
rather than imaginary ones. (You need non-existent ISO settings if
you ask what the "equivalent" on the 30D is for an ISO 100 5D shot.)

The problem with that, of course, is that people rarely take pics of the same scene with both cameras. And even when you do, if the pics are not equivalent, then the FF image will have plusses in one regard (less noise, more center sharpness) and minuses in another (DOF too shallow, for some people, more vignetting, softer edges).

Why not just put the cameras on equivalent settings and show the FF camera stronger all the way through? If you really want to show the pros of FF, take pics in good light, or use a tripod, of scenes where you sacrifice shutter speed for ISO so the FF image has more detail and less noise.

If you shoot two scenes at the same FOV, f-ratio, shutter, and ISO, all you've shown is that the FF camera is capable of more shallow DOF and less noise at the expense of deeper DOF, more vignetting, and softer edges. Those are not plusses to many people, and they thus perpetuate myths about the "shortcomings" of FF, when, in fact, the FF camera did not have to use the same f-ratio, and could have gotten an equivalent image.

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

Please feel free to criticize, make suggestions, and edit my photos. If you wish to use any of my photos for any purpose other than editing in these forums, please ask.

 

[woland21]

woland21 wrote:
If you don't change the lens, the composition changes, but the larger
sensor collects a more light - assuming the image circle is large
enough.

No, please read my very reasonable assumptions again. They are not
complicated, yet they are not at all what you say that they are:
Same aperture
Same FOV
just change the sensor size

Something is wrong here! If you want the same FOV for a different
sensor-sizes, then you have to use another lens with a different
focal length!

Yes, of course. What is wrong with this? just keep the aperture constant.

 

[woland21]

woland21 wrote:
The larger sensor is entirely helpful in gathering more light because
it has more area to capture light. If you want to cripple your
sensor by starving it for light, that's your business.

The larger sensor area does not help in collecting light. A lens is perfectly capable of projecting the same light on a smaller sensor.

The large sensor is better because, as it is used to capture a
photograph at the base ISO, same FOV, same f/stop, and same shutter
speed, i.e., exactly the way nearly everybody would ask the question,
it captures more light.

In this case you can say that the larger sensor is better, if you just want to chose a camera and don't care about what is really going on inside the camera. What is going on is that your assumption of constant f/stop together with a larger sensor is another way of saying that you are using a larger lens. It is the larger lens that really made the difference, not the larger sensor.

I'm just trying to get you to realize how odd these assumptions are.

They are not at all odd. In fact, it is just the assumptions that a camera manufacturer would look at.

It is clearly assumptions that you are not used to, but they are not odd. The fact that photographers are not used to these assumptions is the reason for the whole confusion of believeng that "a larger pixel gathers more light".

No they are not, you just have to adjust the focal length to the
changing sensor size. My assumptions are perfectly valid. You just
have to read them.

Apparently you mean something different by "size of lens" than other
folks. Care to explain how you're changing the focal length but
keeping the size of the lens constant?

I have many times in this thread clearly stated that by size I mean aperture. Aperture is the only size aspect of a lens that you cannot do anything about, no matter how many clever lenses you put inside it. If you want a large aperture you must make a large lens.

 

[woland21]

I think I see at least one source of confusion. When you say "size
of lens" I would guess that you actually mean "(physical) size of
aperture".

You dont need to guess this, I must have written it ten times in this thread.

Your version is still logically inconsistent in places,

No, it isn't.

but if we
stick with the version where you are playing the same game Joe is,

I am not playing any games. I am just saying that a larger lens does not gather more light, if you compare pictures of the same things. I can calculate the light gathering characteristics of any camera on the market without even knowing what size of sensor it has inside. The sensor size makes no difference.

i.e., insisting upon keeping the aperture small even when you scale
up to a longer focal length lens,

Insisting on keeping the aperture CONSTANT while changing the size of the sensor. It doesn't matter if the aperture is large or small. That is the way to understand what is really happening when you change the sensor size.

you are still constructing an
artificial situation where you intentionally starve the large sensor
for light.

I am not starving anything, I am just giving the large sensor the same conditions as the smaller sensor.

So, now you've jumped through all of these hoops to starve your big
sensor for light (by spreading the light that hits a smaller sensor
over a wider area) and you conclude from this allegedly simple, fair,
and natural comparison that big sensors don't capture more light.

Fortunately, there is no need to actually do this experiment. The physics of the situation is already well known, and camera manufacturers undoubtedly knows exactly how this works. Besides, there are much easier ways to vindicate the theory, for those who want, than the way you describe.

In contrast, you have described the experiment where both sensors
receive light at the same intensity, i.e., the case where the images
are shot at the same ISO, same f/stop, same shutter speed, etc. as
"misleading".

This depends on what conclusions you draw from that experiment. If you draw the conclusion that the greater laight gathering capacity of the alrger lens is actually due to the larger sensor, then yes, the conclusions are then misleading.

 

[woland21]

So the
larger sensor has 14 more light falling on a photosite and thus a
higher SNR so less likely to fall below the threshold of noise being
observable).

The reason that your larger sensor has 14 times more light falling on it, which it has in your example, has nothing whatsoever to do with the increased size of the sensor. It is only due to the fact that your assumptions dictates that you use a much larger lens aperture for the larger sensor.

If you used the same lense aperture for the smaller sensor it would get precisely teh same amount of light on its smaller pixels.

 

[woland21]

OK, glad I did not criticize before I fully understood what you were
saying. Large sensor cameras gather more light than small sensor
cameras.

Yes, because they come with larger lenses with larger apertures.

If cameras where sold with lenses of the same apertures, but smaller sensors, then those cameras would gather the same amount of light.

If you want good light gathering performance you should look for cameras with large apertures, and ignore sensor size. It just happens, for other good reasons, that cameras with large apertures always come with large sensors.

So, for the purpose of chosing a camera, it doesn't matter if you have misunderstood the mechanics of the situation and believe that a larger sensor gathers more light. Just follow your incorrect ideas and the camera manufacturers will make sure that you get what you really want, a larger aperture.

(There are however other reasons for a large sensor being better in low light situations than these.)