Small and large sensor cameras with the same lens entrance pupil

TheMartyGuy said:

Suppose we have two 16Mpix cameras. One has an APC sensor and one has a tiny 1/2.3 sensor. Suppose also that the lens on both cameras have the same entrance pupil so that the same number of photons impinge on each pixel of both cameras.

Then why is the high ISO image quality of the APC camera better than that of the 1/2.3 camera?

Click to expand...

Because the larger sensor most likely can absorb more light before oversaturating.

For example, the equivalence ratio (crop factor) between 1/2.3" and APS-C is 3.7x. Let's say 4x to make the math easier.

So, 10mm f/2.8 1/100 ISO 100 on the 1/2.3" sensor camera is equivalent to 40mm f/11 1/100 ISO 1600 on the APS-C camera, since the entrance pupil diameters are the same (10mm / 2.8 = 40mm / 11 = 3.6mm) and the shutter speeds are the same.

But the APS-C camera could shoot the scene instead at 40mm f/11 1/6 ISO 100, motion blur and/or camera shake permitted, thus putting 16x more light on the sensor, resulting in 1/4 the noise for equally efficient sensors. Alternatively, the APS-C camera could shoot the scene at 40mm f/2.8 1/100 ISO 100, again putting 16x more light on the sensor, but this time with 1/4 the DOF.

TheMartyGuy said:

I know from reading that under these circumstances the APC camera is supposed to have better high ISO image quality, but I don't understand why as the same number of photons hits the pixels of each camera.

Click to expand...

For equivalent photos (same DOF and shutter speed), all systems put the same total amount of light on the sensor. The advantage of larger sensor systems is when they can use longer shutter speeds at base ISO and/or wider apertures (entrance pupils) at higher ISOs (necessarily resulting in a more shallow DOF).

TheMartyGuy said:

Thanks for helping me sort this out.

Click to expand...

Hope this helps.

TheMartyGuy said:

Suppose we have two 16Mpix cameras. One has an APC sensor and one has a tiny 1/2.3 sensor. Suppose also that the lens on both cameras have the same entrance pupil so that the same number of photons impinge on each pixel of both cameras.

Then why is the high ISO image quality of the APC camera better than that of the 1/2.3 camera?

I know from reading that under these circumstances the APC camera is supposed to have better high ISO image quality, but I don't understand why as the same number of photons hits the pixels of each camera.

Thanks for helping me sort this out.

Click to expand...

With the same pupil, angle of view and shutter speed, the theoretical image is the same.

But with current sensor technology the saturation level of the pixel is proportional to the area of the pixel. That means the APS-C sensor can count ~16x more photons than the small sensor. (Photon/electron well depth - each photon creates a free electron if it is absorbed, and the pixel acts like a capacitor, where more area means more capacity) If they have the same per pixel read noise, then 16x more photons is 4 stops more dynamic range. At equivalent settings in terms of aperture pupil and shutter speed, this would translate into much better resistance to blowing out highlights.

TheMartyGuy said:

TheMartyGuy wrote:
so that the same number of photons impinge on each pixel of both cameras.

Click to expand...

My less-than-informed impression is that the per-pixel noise is higher with smaller sensors than larger sensors (assuming equivalent technology, filters, software, etc.)

Or not?

Ember42 said:

But with current sensor technology the saturation level of the pixel is proportional to the area of the pixel.

Click to expand...

Not so much a technology issue as a design one. Small pixels could be designed with a much higher saturation level, but in practice for photographic use they are designed with a saturation level that suits a base ISO of 100ISO or higher

l_d_allan said:

l_d_allan said:

TheMartyGuy wrote:
so that the same number of photons impinge on each pixel of both cameras.

Click to expand...

My less-than-informed impression is that the per-pixel noise is higher with smaller sensors than larger sensors (assuming equivalent technology, filters, software, etc.)

Or not?

Click to expand...

Are you assuming same pixel size, or are you assuming same pixel count?

Are you assuming same exposure (same f-stop and shutter speed), or are you assuming same total amount of light (same lens entrance pupil and shutter speed)?

This discussion has really been informative. I now understand that even if the number of photons hitting each pixel is the same for both the large and small sensor, the small sensor pixel will saturate earlier causing blowout of highlights. Each pixel can only hold so many photons and the smaller the pixel real estate the fewer photons the pixel can hold.

TheMartyGuy said:

This discussion has really been informative. I now understand that even if the number of photons hitting each pixel is the same for both the large and small sensor, the small sensor pixel will saturate earlier causing blowout of highlights. Each pixel can only hold so many photons and the smaller the pixel real estate the fewer photons the pixel can hold.

Click to expand...

Unless of course the small pixel designers had made their sensor saturate with more photons, by increasing the internal capacitance for instance.

Great Bustard said:

So, 10mm f/2.8 1/100 ISO 100 on the 1/2.3" sensor camera is equivalent to 40mm f/11 1/100 ISO 1600 on the APS-C camera, since the entrance pupil diameters are the same (10mm / 2.8 = 40mm / 11 = 3.6mm) and the shutter speeds are the same.

Click to expand...

In this exact example, will these two images have the same dynamic range?

--

J.V.

That is very interesting. Has anyone ever made a small sensor with large capacitance so that its performance is similar to a large sensor?

For example, has anyone ever made a 1/2.5 sensor with large enough capacitance to rival an APC sensor in dynamic range?

TheMartyGuy said:

That is very interesting. Has anyone ever made a small sensor with large capacitance so that its performance is similar to a large sensor?

Click to expand...

The Aptina sensors used for the Nikon 1 series have a switchable capacitance (that is a capacitor in the pixel that can be switched in and out). The downside of a high capacitance (and by implication, large pixel) is that it reduces conversion gain and thereby increases input referred read noise. The switchable idea gives them the best of both worlds, although they still put the base ISO at 100. Really it should be more like 40 to maintain absolute parity with FF.

TheMartyGuy said:

For example, has anyone ever made a 1/2.5 sensor with large enough capacitance to rival an APC sensor in dynamic range?

Click to expand...

Not that I know of. Many of the 1/2.4 cameras have a base ISO of 80 rather than 100, but really they need a base ISO of 20 or so.

jvkelley said:

Great Bustard said:

So, 10mm f/2.8 1/100 ISO 100 on the 1/2.3" sensor camera is equivalent to 40mm f/11 1/100 ISO 1600 on the APS-C camera, since the entrance pupil diameters are the same (10mm / 2.8 = 40mm / 11 = 3.6mm) and the shutter speeds are the same.

Click to expand...

In this exact example, will these two images have the same dynamic range?

Click to expand...

If the sensors are equally efficient, yes. However, newer tech often comes out on smaller sensors before larger sensors (e.g. BSI).

l_d_allan said:

l_d_allan said:

TheMartyGuy wrote:
so that the same number of photons impinge on each pixel of both cameras.

Click to expand...

My less-than-informed impression is that the per-pixel noise is higher with smaller sensors than larger sensors (assuming equivalent technology, filters, software, etc.)

Or not?

Click to expand...

...except inasmuch as the pixels contribute to the overall noise in a photo. For example, consider two sensors of the same size, but Sensor A has four times as many pixels as Sensor B. Then, in terms of how noisy the photo looks, we'd compare four pixels from Sensor A to one pixel from Sensor B.

On the other hand, let's say that two sensors had the same pixel count, but Sensor A is twice as large (four times the area) as Sensor B. Then it would be meaningful to compare pixel to pixel, and a pixel from Sensor A would gather 4x as much light as a pixel from Sensor B, and thus have half the noise for a given exposure (but the same noise for the same DOF and shutter speed), assuming the same read noise per pixel.

bobn2 said:

TheMartyGuy said:

That is very interesting. Has anyone ever made a small sensor with large capacitance so that its performance is similar to a large sensor?

Click to expand...

The Aptina sensors used for the Nikon 1 series have a switchable capacitance (that is a capacitor in the pixel that can be switched in and out). The downside of a high capacitance (and by implication, large pixel) is that it reduces conversion gain and thereby increases input referred read noise. The switchable idea gives them the best of both worlds, although they still put the base ISO at 100. Really it should be more like 40 to maintain absolute parity with FF.

bobn2 said:

For example, has anyone ever made a 1/2.5 sensor with large enough capacitance to rival an APC sensor in dynamic range?

Click to expand...

Not that I know of. Many of the 1/2.4 cameras have a base ISO of 80 rather than 100, but really they need a base ISO of 20 or so.

Click to expand...

Interesting!

You learn something new every day.

But ... of course ... that is how it is.

A higher capacitance will increase the bucket depth, but it takes longer to fill it, so the base ISO goes down.

But, what happens if you expose a base ISO 100 at ISO 100 or a base ISO 20 at ISO 100? Any disadvantages for the latter?

Roland Karlsson said:

bobn2 said:

TheMartyGuy said:

That is very interesting. Has anyone ever made a small sensor with large capacitance so that its performance is similar to a large sensor?

Click to expand...

The Aptina sensors used for the Nikon 1 series have a switchable capacitance (that is a capacitor in the pixel that can be switched in and out). The downside of a high capacitance (and by implication, large pixel) is that it reduces conversion gain and thereby increases input referred read noise. The switchable idea gives them the best of both worlds, although they still put the base ISO at 100. Really it should be more like 40 to maintain absolute parity with FF.

Roland Karlsson said:

For example, has anyone ever made a 1/2.5 sensor with large enough capacitance to rival an APC sensor in dynamic range?

Click to expand...

Not that I know of. Many of the 1/2.4 cameras have a base ISO of 80 rather than 100, but really they need a base ISO of 20 or so.

Click to expand...

Interesting!

You learn something new every day.

But ... of course ... that is how it is.

A higher capacitance will increase the bucket depth, but it takes longer to fill it, so the base ISO goes down.

But, what happens if you expose a base ISO 100 at ISO 100 or a base ISO 20 at ISO 100? Any disadvantages for the latter?

Click to expand...

Then we come to the 'ISOless' think. If your readout chain needs to boost gain to overcome ADC noise with low exposures then base 20 and 100 would be a bit like exposing base 100 at 500 on an ISOful camera. If the camera is ISOless, no problem. Except the capacitance gives you big pixel read noise, hence the switchable scheme in the Aptina sensor.

l_d_allan said:

l_d_allan said:

TheMartyGuy wrote:
so that the same number of photons impinge on each pixel of both cameras.

Click to expand...

My less-than-informed impression is that the per-pixel noise is higher with smaller sensors than larger sensors (assuming equivalent technology, filters, software, etc.)

Or not?

Click to expand...

In practice yes, but that is at least partially because you almost never fulfill the constraints of the OP's scenario - same field of view with the same actual aperture. The small sensor almost never has as many photons per image as the larger sensor.

Great Bustard said:

jvkelley said:

Great Bustard said:

So, 10mm f/2.8 1/100 ISO 100 on the 1/2.3" sensor camera is equivalent to 40mm f/11 1/100 ISO 1600 on the APS-C camera, since the entrance pupil diameters are the same (10mm / 2.8 = 40mm / 11 = 3.6mm) and the shutter speeds are the same.

Click to expand...

In this exact example, will these two images have the same dynamic range?

Click to expand...

If the sensors are equally efficient, yes. However, newer tech often comes out on smaller sensors before larger sensors (e.g. BSI).

Click to expand...

Yes, this mitigates the effect somewhat. If new tech means more fatal flaws per unit area, which is catastrophic for larger sensors. In our example above if the yield of the small sensor is 90% (for this kind fo flaw), then the yield of the larger sensor is only 18%... Given the extra die area for the larger sensor, this would imply that the larger sensor should cost ~80x as much as the small sensor!

Ember42 said:

Great Bustard said:

jvkelley said:

Great Bustard said:

So, 10mm f/2.8 1/100 ISO 100 on the 1/2.3" sensor camera is equivalent to 40mm f/11 1/100 ISO 1600 on the APS-C camera, since the entrance pupil diameters are the same (10mm / 2.8 = 40mm / 11 = 3.6mm) and the shutter speeds are the same.

Click to expand...

In this exact example, will these two images have the same dynamic range?

Click to expand...

If the sensors are equally efficient, yes. However, newer tech often comes out on smaller sensors before larger sensors (e.g. BSI).

Click to expand...

Yes, this mitigates the effect somewhat. If new tech means more fatal flaws per unit area, which is catastrophic for larger sensors. In our example above if the yield of the small sensor is 90% (for this kind fo flaw), then the yield of the larger sensor is only 18%... Given the extra die area for the larger sensor, this would imply that the larger sensor should cost ~80x as much as the small sensor!

Click to expand...

This is why the new tech often comes in smaller sensors first (in fact, the cell phone camera sensors are likely driving the consumer tech). However, there's also the matter of read noise vs pixel size. If smaller pixels had less read noise as a general rule, then smaller sensors would be more efficient than larger sensors for equivalent photos (same DOF and shutter speed) as a general rule.

Ember42 said:

l_d_allan said:

Ember42 said:

TheMartyGuy wrote:
so that the same number of photons impinge on each pixel of both cameras.

Click to expand...

My less-than-informed impression is that the per-pixel noise is higher with smaller sensors than larger sensors (assuming equivalent technology, filters, software, etc.)

Or not?

Click to expand...

In practice yes, but that is at least partially because you almost never fulfill the constraints of the OP's scenario - same field of view with the same actual aperture. The small sensor almost never has as many photons per image as the larger sensor.

Click to expand...

This scenario will occur for the same DOF and shutter speed.

Great Bustard said:

TheMartyGuy said:

Suppose we have two 16Mpix cameras. One has an APC sensor and one has a tiny 1/2.3 sensor. Suppose also that the lens on both cameras have the same entrance pupil so that the same number of photons impinge on each pixel of both cameras.

Then why is the high ISO image quality of the APC camera better than that of the 1/2.3 camera?

Click to expand...

Because the larger sensor most likely can absorb more light before oversaturating.

For example, the equivalence ratio (crop factor) between 1/2.3" and APS-C is 3.7x. Let's say 4x to make the math easier.

So, 10mm f/2.8 1/100 ISO 100 on the 1/2.3" sensor camera is equivalent to 40mm f/11 1/100 ISO 1600 on the APS-C camera, since the entrance pupil diameters are the same (10mm / 2.8 = 40mm / 11 = 3.6mm) and the shutter speeds are the same.

But the APS-C camera could shoot the scene instead at 40mm f/11 1/6 ISO 100, motion blur and/or camera shake permitted, thus putting 16x more light on the sensor, resulting in 1/4 the noise for equally efficient sensors. Alternatively, the APS-C camera could shoot the scene at 40mm f/2.8 1/100 ISO 100, again putting 16x more light on the sensor, but this time with 1/4 the DOF.

Click to expand...

Here's a question that came up in another forum: keeping the sensor the same size, why for example wouldn't 100mm f/16 show the same amount of noise as 35mm f/5.6 both on, say, FF? Entrance pupil is the same with both (6.25mm.)

.