Dynamic range FF cameras

Herman Dijkhuis said:

It's claimed by Hassy.

http://www.hasselblad.com/x1d

The X1D can capture up to 14 stops of dynamic range, allowing for unprecedented detail – from the deepest shadows to the brightest highlights. (acc. Hassy)

Click to expand...

They may not measure/define it the same way DXO do.

Charlie Le Denn said:

Interesting: Nikon D810 (@14.8) outperforms the new Hasselblad X1D-50c (@14).

Click to expand...

Not only those numbers are measured using different protocols, they also are far from the practical. Nikon D810 allows to achieve 11, maybe 11.3 stops of photographically useful DR, X1D ... well, close to 12.

To extract the maximum possible DR one needs to expose as hot as possible, and that is the most important skill here.

The DxoMark measurements, linked in another post , are where many people go.
But there are some technical reasons to think they're criteria is flawed.
I also test sensors including testing for dynamic range.
DxOMark and I don't test exactly the same camera models and our scales are different.

Here's a comparison of the top cameras according to DxOMark on the left and according to my tests (values from the sort-able table below my interactive PDR chart) on the right:



Note that the DxOMark criteria places the APS-C D7200 at #2, ahead of several excellent Full Frame cameras. Although it is a strong performer the #13 rank in my list makes more sense.

--
Bill ( Your trusted source for independent sensor data at http://www.photonstophotos.net )

Iliah Borg said:

Iliah Borg said:

Interesting: Nikon D810 (@14.8) outperforms the new Hasselblad X1D-50c (@14).

Click to expand...

Not only those numbers are measured using different protocols, they also are far from the practical. Nikon D810 allows to achieve 11, maybe 11.3 stops of photographically useful DR, X1D ... well, close to 12.

Click to expand...

Yes, agreed. I get 11.53 for the D810. (I have no X1D-50c figures)

Iliah Borg said:

To extract the maximum possible DR one needs to expose as hot as possible, and that is the most important skill here.

Click to expand...

Yes, for sure.

The Davinator said:

Mark Scott Abeln said:

The limiting factor these days are the lenses. From what I've read, you can basically get at most 11 stops of dynamic range out of even the very best lenses.

Click to expand...

As I've seen film test wedges and densitometer testing out to 18 stops....lenses are capable of well more than 11.

Click to expand...

The DR of a RAW has nothing to do with a lower limit on recording signal. It is about noise. More DR = less read noise in the shadows, always a good thing, regardless of the intensity of the blanket of light.

bclaff said:

Here's a comparison of the top cameras according to DxOMark on the left and according to my tests (values from the sort-able table below my interactive PDR chart) on the right:

Click to expand...

Interesting that your measurements place the A7 before the D810. All else equal (all recent Sony FF sensors), the ISO 64 of the D810 should trump the ISO 100 FF cameras.

Herman Dijkhuis said:

It's claimed by Hassy.

http://www.hasselblad.com/x1d

The X1D can capture up to 14 stops of dynamic range, allowing for unprecedented detail – from the deepest shadows to the brightest highlights. (acc. Hassy)

Click to expand...

Then that's an apple compared to other DR standards' oranges.

DR is not concrete like a yardstick with the first inch or two missing. It's abstract like a yardstick that is chewed by a dog on the low number end, and DR ratings all vary in how chewed is too chewed.

noirdesir said:

Interesting that your measurements place the A7 before the D810. All else equal (all recent Sony FF sensors), the ISO 64 of the D810 should trump the ISO 100 FF cameras.

Click to expand...

The problem is that ISO is not a precise measure. The International Standards Organization's standard 12232 provides of a number of ways to determine it.

Mark Scott Abeln said:

noirdesir said:

Interesting that your measurements place the A7 before the D810. All else equal (all recent Sony FF sensors), the ISO 64 of the D810 should trump the ISO 100 FF cameras.

Click to expand...

The problem is that ISO is not a precise measure. The International Standards Organization's standard 12232 provides of a number of ways to determine it.

Click to expand...

Sure, but the saturation-based ISO (which is quite appropriate for DR measurements), is reported to be ISO 73 for the A7 and ISO 47 for the D810 by DxO, ie, almost two-thirds as a stop lower on the D810. All else equal (eg, the illumination level which marks the lower end of Bill's DR bracket, ie, the same read noise), adding two-thirds of a stop to the top should result in two-thirds of a stop more of a DR range.

sportyaccordy said:

All of them

Yes, even Canon's latest offerings. At least compared to film's ~5-6 stops.

Click to expand...

When we speak film DR, there is 2 different categories.

1) Editing room in scanned file.

2) DR recorded on film.

Because the recorded DR on film is 20EV, there is no need to push scanned file ~5-6 stops. Basically it's ready to print with no editing.

Sugasmune said:

Because the recorded DR on film is 20EV, [...]

Click to expand...

How do you define DR on film and then how do you measure it?

J A C S said:

Sugasmune said:

Because the recorded DR on film is 20EV, [...]

Click to expand...

How do you define DR on film and then how do you measure it?

Click to expand...

Start from 2:07, press CC for subs


--
http://www.libraw.org/

Iliah Borg said:

J A C S said:

Sugasmune said:

Because the recorded DR on film is 20EV, [...]

Click to expand...

How do you define DR on film and then how do you measure it?

Click to expand...

Start from 2:07, press CC for subs

Click to expand...

LOL! I do not need CC.

noirdesir said:

Sure, but the saturation-based ISO (which is quite appropriate for DR measurements), is reported to be ISO 73 for the A7 and ISO 47 for the D810 by DxO, ie, almost two-thirds as a stop lower on the D810. All else equal (eg, the illumination level which marks the lower end of Bill's DR bracket, ie, the same read noise), adding two-thirds of a stop to the top should result in two-thirds of a stop more of a DR range.

Click to expand...

The only think I can think of right now is that the color filter arrays of the cameras are different. This will have an effect on sensitivity.

Mark Scott Abeln said:

noirdesir said:

Sure, but the saturation-based ISO (which is quite appropriate for DR measurements), is reported to be ISO 73 for the A7 and ISO 47 for the D810 by DxO, ie, almost two-thirds as a stop lower on the D810. All else equal (eg, the illumination level which marks the lower end of Bill's DR bracket, ie, the same read noise), adding two-thirds of a stop to the top should result in two-thirds of a stop more of a DR range.

Click to expand...

The only think I can think of right now is that the color filter arrays of the cameras are different. This will have an effect on sensitivity.

Click to expand...

I don't think so. If the CFA were denser (ie, let less light through resulting in lower base ISO on the D810 compared to the A7 or the D800), the low light performance would be two-thirds of a stop worse (which it isn't).

No, the most straightforward solution is more read noise in the D810 (but that is not what all other tests of the sensor show), which makes this still a puzzle for me. If anybody than Bill himself is probably best placed to explain the reason.

noirdesir said:

bclaff said:

Here's a comparison of the top cameras according to DxOMark on the left and according to my tests (values from the sort-able table below my interactive PDR chart) on the right:

Click to expand...

Interesting that your measurements place the A7 before the D810. All else equal (all recent Sony FF sensors), the ISO 64 of the D810 should trump the ISO 100 FF cameras.

Click to expand...

But things aren't equal; they don't appear to be identical technology and the pixel pitch is different.
The smaller D810 pixels mean that the normalized performance is better.

A7R FWC measures as about 48-50K, D810 - as about 75-80K.

Just a few points re DR:

When people talk about dynamic range they're not necessarily always meaning the same thing. Many people are specifically referring to the amount of noise visible when bringing up shadows. See dpr articles on ISO invariance and decide if this is what's important to you.

DR scores read across different sites cannot be used for comparison.

DXO is a company that sells image editing software. They publish their sensor testing data, which many find interesting, but they're not actually a standards authority.

jpeg DR is irrelevant. That just represents a manufacturer's choice about how "punchy" they want the default to look.

A camera doesn't have one single DR. It has a set of DR values at different ISOs. Typically Canon cameras have poorer DR at base ISO but often catch up at higher ISOs. So if you're buying a FF camera in the expectation that you'll be shooting at ISO3200 forget about that 14.8 EV of the Nikon D810.

If you're shooting things with no movement then the Pentax K1 pixel shift mode captures more light and will beat anything else.

They're all good, and the number of situations where it's possible to get a good shot with one but not possible, by any means, with another are few.

Bill Claff is the guy who actually knows about this stuff, so search out his comments.

John Sheehy said:

Herman Dijkhuis said:

It's claimed by Hassy.

http://www.hasselblad.com/x1d

The X1D can capture up to 14 stops of dynamic range, allowing for unprecedented detail – from the deepest shadows to the brightest highlights. (acc. Hassy)

Click to expand...

Then that's an apple compared to other DR standards' oranges.

DR is not concrete like a yardstick with the first inch or two missing. It's abstract like a yardstick that is chewed by a dog on the low number end, and DR ratings all vary in how chewed is too chewed.

Click to expand...

Very good analogy.

whakapu said:

A camera doesn't have one single DR. It has a set of DR values at different ISOs. Typically Canon cameras have poorer DR at base ISO but often catch up at higher ISOs. So if you're buying a FF camera in the expectation that you'll be shooting at ISO3200 forget about that 14.8 EV of the Nikon D810.

If you're shooting things with no movement then the Pentax K1 pixel shift mode captures more light and will beat anything else.

Click to expand...

I think that with still subjects, any method of taking several shots and averaging them will reduce noise and hence increase DR at the low end. The astrophotographers use this technique regularly, and Sigma have a new setting which they call SFD using a similar approach to good effect.

whakapu said:

They're all good, and the number of situations where it's possible to get a good shot with one but not possible, by any means, with another are few.

Bill Claff is the guy who actually knows about this stuff, so search out his comments.

Click to expand...