The Most Useful ISO, etc. Graph

[Michael Erlewine]

This graph allows you to compare and see the differences between dozens of cameras, including the Nikon D810, Hasselblad X1D, GFX 50s, etc.

I have no idea how accurate, but for the cameras I know, it seems right on the money. It shows how close the Nikon D810 is to the new medium-format mirrorless camers.

http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%201D%20X%20Mark%20II,Nikon%20D5,Sony%20ILCE-9,Sony%20ILCE-9(ES)

[Akira]

Well, D750 gets even closer:

http://www.photonstophotos.net/Charts/PDR.htm#FujiFilm%20GFX%2050S,Hasselblad%20X1D-50c,Nikon%20D750,Nikon%20D810

That said, I noticed that the performance datum for D750 had become noticeably better after the slight change of the design of the chart.  All the other data for all the other models seem to have remained the same.  The data for this "Photons to Photos" website refers to the measurements of DXOmarks, and the data for D750 has remained the same.

I don't know why.

[Akira]

FWIW, Phase One IQ3 tops the crowd.

http://www.photonstophotos.net/Charts/PDR.htm#FujiFilm%20GFX%2050S,Hasselblad%20X1D-50c,Nikon%20D750,Nikon%20D810,Phase%20One%20IQ3%20100MP

[KenP]

Michael,

I am just starting to learn all about digital photography and a bit of the science behind the sensors. When plugging my two recent and first digital purchases (Nikon 1V1 and D2h) into the graph, the results are rather abysmal. I do wonder, however, how these graphs and charts play out in the actual use of the camera. You suggest based on your experience, the chart seems accurate.

How do you test for such a result? Size of a print? I am still just learning to use my camera but wonder what kind of results I could get in lower light. Particularly for shooting my kids sports which is about the only chance I get to use a camera thus far. How far can you push the ISO before the results become unacceptable? How "clean" does an image have to be? Is the old film grain of Ilford 3200 no longer desirable?

My biggest challenge thus far has been to attain a high enough shutter speed to capture action when the light starts to fade. I do not have a zoom lens for the 1V1 yet but looking to purchase a 30-110. Still on the slower side at F3.8 but with VR, perhaps it will allow me to push the ISO?

Ken

[David H. Hartman]

Ken,

I suggest thinking of VR as a handy substitute for a monopod. Neither a tripod nor a monopod nor VR can help with subject movement but they all help with camera movement introduce by the photographer. To deal with subject movement a higher shutter speed and or panning with the subject is needed.

Dave

[KenP]

Thanks Dave. One variable to take out of the equation. I now understand VR 

[armando_m]

Ken,
the image does not "need" to be clean,
I have a v1 and it can be great at high iso in black and white, as I feel it does give the look of film

but high iso color images is not the best option, iso400 is about as high as go with the v1 when shooting in color, this is close to the iso 500 shown in web page

[Jack Dahlgren]

How do you test for such a result? Size of a print? I am still just learning to use my camera but wonder what kind of results I could get in lower light. Particularly for shooting my kids sports which is about the only chance I get to use a camera thus far. How far can you push the ISO before the results become unacceptable? How "clean" does an image have to be? Is the old film grain of Ilford 3200 no longer desirable?

You can test for your own purposes and perception. That will be more useful than a chart. Try it out at different ISO's and see what happens. YOU decide what is acceptable.

My old D200 was OK up to about 800 ISO. My Df is decent up to 10,000 ISO. Modern digital cameras are amazing in low light.

[Michael Erlewine]

Well, D750 gets even closer:

Yes, but I want more than 24 Mpx and I like the ISO 64 of the D810.

[Akira]

Yes, but I want more than 24 Mpx and I like the ISO 64 of the D810.

Michael, yes, I know.   

One thing I noticed about the 50MP 44x33mm sensor made by Sony when it started to be used on the digital MF cameras like Phase One IQ3 50MP (which is the same one as used in Hassy X1D, Pentax 645Z and Fuji GFX50S) was that the sensor yield a 14bit file, unlike the conventional MF sensors.  Even though IQ3 50MP creates 16bit RAW files, they are upsampled from the 14bit file, which may limit the dynamic range.  According to the spec sheet of Phase One, the DR of IQ3 100Mp is 15 stops whereas that of 50MP is 14 stops.

[Michael Erlewine]

Michael, yes, I know.   

One thing I noticed about the 50MP 44x33mm sensor made by Sony when it started to be used on the digital MF cameras like Phase One IQ3 50MP (which is the same one as used in Hassy X1D, Pentax 645Z and Fuji GFX50S) was that the sensor yield a 14bit file, unlike the conventional MF sensors.  Even though IQ3 50MP creates 16bit RAW files, they are upsampled from the 14bit file, which may limit the dynamic range.  According to the spec sheet of Phase One, the DR of IQ3 100Mp is 15 stops whereas that of 50MP is 14 stops.

I noticed all that too. Hopefully, Nikon won't screw up the D820 (D850) when it comes to low ISOs. Otherwise, I am out of there and looking to a big Mpx camera from Sony, even though they fiddle with the sensor too much for me.

[bclaff]

Well, D750 gets even closer:

http://www.photonstophotos.net/Charts/PDR.htm#FujiFilm%20GFX%2050S,Hasselblad%20X1D-50c,Nikon%20D750,Nikon%20D810

That said, I noticed that the performance datum for D750 had become noticeably better after the slight change of the design of the chart.  All the other data for all the other models seem to have remained the same.

That's very observant. The data for all cameras was updated and most rose a little bit.
Also, I do update results as I get better data. The D750 is on my list to request a new batch of files. Normally these curves are smoother.

The data for this "Photons to Photos" website refers to the measurements of DXOmarks, and the data for D750 has remained the same.

If you're referring to the second section of PhotonsToPhotos with DxOMark derived data, that is correct.
Since DxOMark never retests a camera curves there would never change.

[bobfriedman]

i've seen this curve many times and i still don't believe the D5 results..

[bclaff]

You can test for your own purposes and perception. That will be more useful than a chart. Try it out at different ISO's and see what happens. YOU decide what is acceptable.

My old D200 was OK up to about 800 ISO. My Df is decent up to 10,000 ISO. Modern digital cameras are amazing in low light.

Your observation lines up perfectly with the chart.
Your personal tolerance for Image Quality (IQ) is a PDR of at least 5.2
This is ISO 800 for the D200 and ISO 10000 for the Df

[Bill De Jager]

How do you test for such a result? Size of a print?

There will be a dynamic range associated with the data in any given digital photograph.  This will vary not just by camera model but by the settings used for taking that photograph, the lighting, etc.  There is a separate dynamic range associated with each output medium, such as color print, black and white print, or computer monitor.  This will also vary by type or model.  For any given photograph, the actual dynamic range you can see will be the lower of these two.  Prints generally have a lower dynamic range than other common display media.  That means they are best used where needed for convenience, not for getting the best view of a photograph.

Size of a print does relate to resolution.  A photo with lower resolution, whether film or digital, can be enlarged less on a print or monitor before it loses apparent sharpness.  But this is a separate matter from dynamic range.