Why exactly do large sensors have more dynamic range

Started 5 months ago | Discussions
HDRI Forum Member • Posts: 85
Why exactly do large sensors have more dynamic range

Many resources(websites) attribute the large sensors having more DR due to they having bigger pixels. But in current days , GFX100 has 3.75 µm pixel pitch and Sony A7Siii has 8.4 µm pixel pitch. And yet GFX100 has higher PDR than Sony by around 1.5 stops.

So what exactly makes larger sensors have more DR.

Or if a company wishes they can actually have FF sensor like A7Siii have more/same DR as GFX 100 but they don't so for some other marketing reasons ?

 HDRI's gear list:HDRI's gear list
Canon EOS 350D Fujifilm X-T100 Canon EF 100mm f/2.0 USM Canon 70-300 F4-5.6 IS II Fujifilm 15-45mm F3.5-5.6 OIS PZ
JimKasson
MOD JimKasson Forum Pro • Posts: 34,793
Re: Why exactly do large sensors have more dynamic range
4

HDRI wrote:

Many resources(websites) attribute the large sensors having more DR due to they having bigger pixels. But in current days , GFX100 has 3.75 µm pixel pitch and Sony A7Siii has 8.4 µm pixel pitch. And yet GFX100 has higher PDR than Sony by around 1.5 stops.

So what exactly makes larger sensors have more DR.

Or if a company wishes they can actually have FF sensor like A7Siii have more/same DR as GFX 100 but they don't so for some other marketing reasons ?

You're talking per-image DR, not per-pixel DR, right? Useful base-ISO DR these days is pretty much controlled by shot noise, not read noise. Signal to noise ratio of shot noise goes as the square root of the number of electrons counted. Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

The only useful way at present to get more FWC across the entire sensor is to make the sensor bigger.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
SrMi
SrMi Senior Member • Posts: 2,199
Re: Why exactly do large sensors have more dynamic range

HDRI wrote:

Many resources(websites) attribute the large sensors having more DR due to they having bigger pixels. But in current days , GFX100 has 3.75 µm pixel pitch and Sony A7Siii has 8.4 µm pixel pitch. And yet GFX100 has higher PDR than Sony by around 1.5 stops.

Those resources are obviously wrong.

So what exactly makes larger sensors have more DR.

See Jim's answer.

Or if a company wishes they can actually have FF sensor like A7Siii have more/same DR as GFX 100 but they don't so for some other marketing reasons ?

JimKasson
MOD JimKasson Forum Pro • Posts: 34,793
Re: Why exactly do large sensors have more dynamic range
2

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
Dan Wells Contributing Member • Posts: 584
Re: Why exactly do large sensors have more dynamic range

Why would the A7r IV have a lower per unit area FWC than the GFX? I thought those two sensors were very, very closely related... I certainly see Jim's point about per-sensor FWC, but the per-area FWC shouldn't differ by 10% as far as I can see...

It also doesn't surprise me that the Z7 is unusually high per unit area - it has the low ISO settings - would the sensor modifications that enable ISO 64 also increase FWC?

Erik Kaffehr
Erik Kaffehr Veteran Member • Posts: 6,068
Re: Why exactly do large sensors have more dynamic range

Dan Wells wrote:

Why would the A7r IV have a lower per unit area FWC than the GFX? I thought those two sensors were very, very closely related... I certainly see Jim's point about per-sensor FWC, but the per-area FWC shouldn't differ by 10% as far as I can see...

It also doesn't surprise me that the Z7 is unusually high per unit area - it has the low ISO settings - would the sensor modifications that enable ISO 64 also increase FWC?

One theory is that Nikon uses a bit of the non-linear part of the photon transfer curve.

Best regards

Erik

-- hide signature --

Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

JimKasson
MOD JimKasson Forum Pro • Posts: 34,793
Re: Why exactly do large sensors have more dynamic range

Dan Wells wrote:

Why would the A7r IV have a lower per unit area FWC than the GFX?

Experimental error in my measurements? Different threshold nonlinearity? Sony’s calibration prescaling? All of the above?

I thought those two sensors were very, very closely related... I certainly see Jim's point about per-sensor FWC, but the per-area FWC shouldn't differ by 10% as far as I can see...

It also doesn't surprise me that the Z7 is unusually high per unit area - it has the low ISO settings - would the sensor modifications that enable ISO 64 also increase FWC?

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
MOD Doppler9000 Contributing Member • Posts: 544
Re: Why exactly do large sensors have more dynamic range

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

The QHY600 monochrome astro camera, which uses a monochrome IMX455, has a full well capacity of >51,000 e- > 3,607 e-/um^2. (51/37 x 2617).

I assume the advantage lies in the lack of a CFA.

JimKasson
MOD JimKasson Forum Pro • Posts: 34,793
Re: Why exactly do large sensors have more dynamic range

Doppler9000 wrote:

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

The QHY600 monochrome astro camera, which uses a monochrome IMX455, has a full well capacity of >51,000 e- > 3,607 e-/um^2. (51/37 x 2617).

I assume the advantage lies in the lack of a CFA.

I don’t see how removing the CFA would affect the FWC. Maybe that camera uses a different nonlinearity threshold.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
MOD Doppler9000 Contributing Member • Posts: 544
Re: Why exactly do large sensors have more dynamic range

JimKasson wrote:

I don’t see how removing the CFA would affect the FWC. Maybe that camera uses a different nonlinearity threshold.

Is this spec typically published or is it left less clear?

https://www.qhyccd.com/index.php?m=content&c=index&a=show&catid=94&id=55

Erik Kaffehr
Erik Kaffehr Veteran Member • Posts: 6,068
Re: Why exactly do large sensors have more dynamic range
1

Hi,

Part of the explanation may be the pixel size. DR is decided by FWC (Full Well Capacity) / readout noise.

Readout noise depends much on the voltage created over the pixels. That voltage is the charge in the pixels divided by the capacitance of the pixels. Few electron charges over a large pixel yield low voltage which results in high noise.

The Sony A7sIII has huge pixels and very high readout noise at base ISO (around 14.3 EV), the Sony A7rIV has a readout noise around 2.9EV at base ISO.

From Bill Claff's site.

My guess may be that the A7sIII may be limited by 14 bits readout. FWC seems to be around 225 000 e-. 14 bit readout means around 16000 values, so each data number represents about 16 electron charges.

The A7sIII is intended for video, so it needs fast readout.

Note that the sensor switches to high gain conversion at 1600 ISO and readout noise drops to 1.1 e-.

Check the DR values at DxO-mark.

The per pixel values tell a different story.

Looking at the pixel level, SNR (Signal Noise Ratio) is much better on the A7sIII

But, seeing the whole image it is a bit worse off.

What we can see is that the A7sIII doesn't have enough pixels to carry all information from the sensor. But, it is intended for 4K video, not stills.  Another reason 8K video may make sense.

Best regards

Erik

-- hide signature --

Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

Erik Kaffehr
Erik Kaffehr Veteran Member • Posts: 6,068
Re: Why exactly do large sensors have more dynamic range

Doppler9000 wrote:

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

The QHY600 monochrome astro camera, which uses a monochrome IMX455, has a full well capacity of >51,000 e- > 3,607 e-/um^2. (51/37 x 2617).

I assume the advantage lies in the lack of a CFA.

Using more of the non linear part of the photon transfer curve? That may make some sense on a monochrome device.

Best regards

Erik

-- hide signature --

Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

OP HDRI Forum Member • Posts: 85
Re: Why exactly do large sensors have more dynamic range

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

So if I understand right , the Z7 sensor tech at MF size would yield higher PDR than GFX at base iso ?

 HDRI's gear list:HDRI's gear list
Canon EOS 350D Fujifilm X-T100 Canon EF 100mm f/2.0 USM Canon 70-300 F4-5.6 IS II Fujifilm 15-45mm F3.5-5.6 OIS PZ
Erik Kaffehr
Erik Kaffehr Veteran Member • Posts: 6,068
Re: Why exactly do large sensors have more dynamic range
1

HDRI wrote:

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

So if I understand right , the Z7 sensor tech at MF size would yield higher PDR than GFX at base iso ?

No,

You can deduce that the Z7 has higher SNR at the pixel level.

For PDR, you need to take the readout noise and the pixel count into account. The Z7 has larger pixels which means fewer pixels for any given size.

The easiest way to find out is to check Bill Claffs website and check for the GFX 100 in crooped sensor mode:

Both have maximum PDR around 11.6 EV. The ISO axis is nominal, so the GFX 100 seems to achieve 11.6EV at higher ISO, but that figure is not measured.

https://www.photonstophotos.net/Charts/PDR.htm#FujiFilm%20GFX%20100(FF),Nikon%20Z%207II
This seems a bit like counting the angels on the head of a pin. There are measurable differences, but the major factor is how the photographer exposes.

The winner would be the camera with the best histogram.

Best regards

Erik

-- hide signature --

Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

OP HDRI Forum Member • Posts: 85
Re: Why exactly do large sensors have more dynamic range

Erik Kaffehr wrote:

HDRI wrote:

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

So if I understand right , the Z7 sensor tech at MF size would yield higher PDR than GFX at base iso ?

No,

You can deduce that the Z7 has higher SNR at the pixel level.

For PDR, you need to take the readout noise and the pixel count into account. The Z7 has larger pixels which means fewer pixels for any given size.

Hmm, I thought the rightmost column values are normalised to unit area (hence pixel size independent)

 HDRI's gear list:HDRI's gear list
Canon EOS 350D Fujifilm X-T100 Canon EF 100mm f/2.0 USM Canon 70-300 F4-5.6 IS II Fujifilm 15-45mm F3.5-5.6 OIS PZ
Erik Kaffehr
Erik Kaffehr Veteran Member • Posts: 6,068
Re: Why exactly do large sensors have more dynamic range

HDRI wrote:

Erik Kaffehr wrote:

HDRI wrote:

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

So if I understand right , the Z7 sensor tech at MF size would yield higher PDR than GFX at base iso ?

No,

You can deduce that the Z7 has higher SNR at the pixel level.

For PDR, you need to take the readout noise and the pixel count into account. The Z7 has larger pixels which means fewer pixels for any given size.

Hmm, I thought the rightmost column values are normalised to unit area (hence pixel size independent)

Yes, but FWC is the nominator of the equation. The denominator is a bit different as readout noise adds in quadrature.

To that may come that the sensors have a lot of tunable parameters.

Best regards

Erik

-- hide signature --

Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

JimKasson
MOD JimKasson Forum Pro • Posts: 34,793
Re: Why exactly do large sensors have more dynamic range

Doppler9000 wrote:

JimKasson wrote:

I don’t see how removing the CFA would affect the FWC. Maybe that camera uses a different nonlinearity threshold.

Is this spec typically published or is it left less clear?

https://www.qhyccd.com/index.php?m=content&c=index&a=show&catid=94&id=55

I’ve never seen a regular consumer camera manufacturer spec FWC.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
JimKasson
MOD JimKasson Forum Pro • Posts: 34,793
Re: Why exactly do large sensors have more dynamic range

HDRI wrote:

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

So if I understand right , the Z7 sensor tech at MF size would yield higher PDR than GFX at base iso ?

I think you’re assuming more accuracy to these numbers than is warranted. And besides, although FWC is the long pole in the tent, read noise can’t be entirely ignored.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
MOD Doppler9000 Contributing Member • Posts: 544
Re: Why exactly do large sensors have more dynamic range

Erik Kaffehr wrote:

Doppler9000 wrote:

JimKasson wrote:

JimKasson wrote:

Full well capacity per unit area has maxxed out at around 3000 electrons per square micrometer (3000 e-/um^2).

The QHY600 monochrome astro camera, which uses a monochrome IMX455, has a full well capacity of >51,000 e- > 3,607 e-/um^2. (51/37 x 2617).

I assume the advantage lies in the lack of a CFA.

Using more of the non linear part of the photon transfer curve? That may make some sense on a monochrome device.

Best regards

Erik

I had missed a reference on the product page suggesting that the color version has the same fwc.

For a frame of reference, their IMX411BSI-based 150 mp camera has a rated fwc of > 50,000 e-, with an extended mode with > 80,000 e-.

The 102mp IMX461 camera has a rated fwc of 44,000 e-.

Erik Kaffehr
Erik Kaffehr Veteran Member • Posts: 6,068
Re: Why exactly do large sensors have more dynamic range

JimKasson wrote:

Doppler9000 wrote:

JimKasson wrote:

I don’t see how removing the CFA would affect the FWC. Maybe that camera uses a different nonlinearity threshold.

Is this spec typically published or is it left less clear?

https://www.qhyccd.com/index.php?m=content&c=index&a=show&catid=94&id=55

I’ve never seen a regular consumer camera manufacturer spec FWC.

Sensor manufacturers used to do that, at least some of them. Thinking Kodak spec sheets.

But I don't think FWC is a firm boundary, it may depend on the requirement of linearity.

Best regards

Erik

-- hide signature --

Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

Keyboard shortcuts:
FForum MMy threads