BREAKING: Z7 sensor performance is tested better than D850. Way better resolution & MTF50

[JimKasson]

Why would anyone expect better sensor performance in terms of resolution? This is basically same sensor with new AF sensors, as Bill Claff's noise test has already shown.

Different microlenses?

Why different? We had comments from Marianne in the incidence angle where she said difference in corner microlensrs should be minimal.

I wasn't talking about the angle, but the aperture.

Don’t microlenses affect noise measurements, as they try to maximize light gathering?

They won't affect read noise, for sure. They won't affect photon noise, the way that Bill and I do our testing, which is normalized to full scall at base ISO. In the GFX, the camera is less sensitive than the D5 for the same luminance. Some of that is probably due to microlenses, but some could be due to ISO calibration differences. CIPA gives a lot of wiggle room.

If that is same, not s sign that microlenses are same or almost?

As I said, we don't know that the photon noise normalized to input luminance is the same. That could be affected by CFA pigments as well as microlenses.

Secondly, Nikon didn’t make any claims of sensor differences from D850 to Z7, especially re resolving power, so I think it’s fair to assume they’re pretty much same. If there was any relevant diff they be talking about that, no?

Not sure. I think it's probably best to wait until the tech-oriented testers get the cameras in their hands before drawing any firm conclusions.

Jim

 

[rubank]

Why would anyone expect better sensor performance in terms of resolution? This is basically same sensor with new AF sensors, as Bill Claff's noise test has already shown.

Different microlenses?

Please explain how different microlenses can increase resolution, i.e alter the sampling frequency.

 

[JimKasson]

Why would anyone expect better sensor performance in terms of resolution? This is basically same sensor with new AF sensors, as Bill Claff's noise test has already shown.

Different microlenses?

Please explain how different microlenses can increase resolution, i.e alter the sampling frequency.

They don't alter the sampling frequency, but they can reduce the effect on MTF caused by the pixel aperture being finite, as opposed to the Dirac function of an ideal sampler.

There is a tradeoff involved, as making the pixel aperture smaller without changing the pitch increases the amount of aliasing. Nevertheless, that is what Fuji chose to do with the GFX.

Jack Hogan has a discussion of the effect on MTF by sampling at 100% fill factor:

Loading…

www.strollswithmydog.com

Here's an illustration of making the fill factor larger than 100% and making it blurry:

Loading…

www.dspguide.com

Jim

 

[rhlpetrus]

Found a Nikon Rumors post about a IR interview done in Japan, with three Nikon engineers, and were asked about the two sensors, basically the same for raw and visibly better for jpgs because of Expeed 6. Question how much better, i guess that’s what the Pdnonline test is trying to say?

Well, jpeg resolution is likely related to sharpening, as Jim has already mentioned, so not interesting for my photography, as use raw exclusevely.

 

[paradiddlesix66]

Regarding expertise:

Quote

PDN is a member of theTechnical Image Press Association which has contracted with Image Engineering to perform detailed lab tests of digital cameras. Seehere for a full methodological rundown of how Image Engineering puts cameras through their paces. Full res files of every visual in this review are available to download for your pixel-peeping pleasure here.

The lab that did the tests appears to be quite legit, so no worries from me about their expertise.

It appears, however, that all their tests are based on JPEGs, with mostly default settings. That can be defended as "how most consumers will use the camera" for lower end cameras, but it's not a sufficient test methodology for higher end cameras like the Z7, where most users will not just shoot JPEG at factory defaults.

Finally, they used the S-Line 35mm f/1.8 lens on the Z7, and I'm guessing that accounts for much of the improvement in resolution compared to the D850. That's good in one sense — it seems like an early data point to support the notion that the S-Line optics really are exceptional — but it's not so good for making comparisons to Nikon cameras tested with F-mount lenses (like the D850). If you mount whatever lens they used to test the D850 to the Z7, I'm guessing the MTF figures would be closer to the D850's.

Why would anyone expect better sensor performance in terms of resolution? This is basically same sensor with new AF sensors, as Bill Claff's noise test has already shown.

it would surprise me if it isnt a little better than in the d850. i cant think of any nikon cameras that used "the same sensor" in which the newer camera wasnt better.

ie d600 < d610 < d750. sensor design seems to be a game of small improvements over the life cycle of a particular sensors design.

 

[Joejr]

Well sure, i agree with your premise if your shooting raw, basically the two sensors are the same.what has changed is the expeed 6 processors.And Nik engs are saying that jpegs are noticeable better than 850. So is that what the pdn test is saying? If you shoot raw you should expect similar results as the 850, i f you shoot jpegs they should be better.

 

[JimKasson]

Well sure, i agree with your premise if your shooting raw, basically the two sensors are the same.

We don’t know that — yet.

what has changed is the expeed 6 processors.And Nik engs are saying that jpegs are noticeable better than 850. So is that what the pdn test is saying? If you shoot raw you should expect similar results as the 850, i f you shoot jpegs they should be better.

 

[Joejr]

I guess what i should have said was: Z sensor pertty much on par with 850s, you can expect Nef raw files to have similar quality. IR interview August 28 i think.

 

[ARClark]

So, calling this a sensor performance test is grossly misleading.

So many people want 15 minutes of internet forum fame for being the first to post some article about the latest gear regardless of the quality of the material.

And this was actually the second posting of the article. https://www.dpreview.com/forums/post/61669485

 

[Michael1000]

Thanks for posting the test. It was an interesting read.

Some notables:

1.1 second startup time. The Sony A7III and A7RIII are 0.9 seconds, so comparable. The D850 is 0.2 seconds.

The Z7 will shoot JPEGs at 8.7 fps UNTIL THE CARD FILLS. So if you are a JPEG shooter, I guess you can just hold the button!

4K video dynamic range is 10.0 at low ISO (100) and 9.7 at high ISO (1600). The A7RIII is 8.7 and 8.0. The A7III is 9.0 and 8.0. That's a sizeable advantage for the Z7.

 

[badsheep]

Maybe I am misunderstanding but how is it possible to get more LP/H than half the number of pixels ?

 

[JimKasson]

Maybe I am misunderstanding but how is it possible to get more LP/H than half the number of pixels ?

Here's a little introduction to frequency domain sharpness measurements:

Sharpness

Sharpness is arguably the most important single image quality factor: it determines the amount of detail an image can convey. The image on the upper right illustrates the effects of reduced sharpness (from running Image Processing with one of the Gaussian filters set to 0.7 sigma). Device or...

www.imatest.com

The way the measurements are made is more complex than just counting lines. This is only one of the reasons that I much prefer MTF measurements in cycles/picture height (cy/ph). I think it causes less confusion.

Jim

 

[Jack Hogan]

”103 percent of the theoretical maximum”

I believe the theoretical maximum they are talking about assumes 100% effective fill factor. Slanted edge MTF testing measures pixel aperture. Reducing the pixel effective aperture below that giving 100% fill will give sharper results than an ideal AA-less sensor at 100% fill.

An extreme example of this is the Fuji GFX, which has deliberately small micro lenses, and gives off the charts MTFs.

Right Jim. For reference, here is a thread where we discussed fill factor and effective pixel aperture when the Fuji came out.

Comparing undemosaiced slanted edge MTF of a AA-less BSI sensors at base ISO with the same lens is essentially measuring the characteristics of the microlenses. Not clear to me that's what they did, but that's the high road.

 

[rhlpetrus]

Regarding expertise:

Quote

PDN is a member of theTechnical Image Press Association which has contracted with Image Engineering to perform detailed lab tests of digital cameras. Seehere for a full methodological rundown of how Image Engineering puts cameras through their paces. Full res files of every visual in this review are available to download for your pixel-peeping pleasure here.

The lab that did the tests appears to be quite legit, so no worries from me about their expertise.

It appears, however, that all their tests are based on JPEGs, with mostly default settings. That can be defended as "how most consumers will use the camera" for lower end cameras, but it's not a sufficient test methodology for higher end cameras like the Z7, where most users will not just shoot JPEG at factory defaults.

Finally, they used the S-Line 35mm f/1.8 lens on the Z7, and I'm guessing that accounts for much of the improvement in resolution compared to the D850. That's good in one sense — it seems like an early data point to support the notion that the S-Line optics really are exceptional — but it's not so good for making comparisons to Nikon cameras tested with F-mount lenses (like the D850). If you mount whatever lens they used to test the D850 to the Z7, I'm guessing the MTF figures would be closer to the D850's.

Why would anyone expect better sensor performance in terms of resolution? This is basically same sensor with new AF sensors, as Bill Claff's noise test has already shown.

it would surprise me if it isnt a little better than in the d850. i cant think of any nikon cameras that used "the same sensor" in which the newer camera wasnt better.

ie d600 < d610 < d750. sensor design seems to be a game of small improvements over the life cycle of a particular sensors design.

In terms of resolution?

 

[Azimuth Jones]

JimKasson, post: 61673466, member: 1261637"]
Of course, another way to get MTF's way over what you'd expect from a perfect lens on a 100% fill factor sensor is to sharpen the image. And that includes the stealth sharpening performed by some raw developers and most in-camera-JPEGs.

Jim, they tested JPEGs at default sharpening settings.

Then the numbers are pretty useless for judging how sharp raw files would be.
[/QUOTE]
True. But who was claiming jpeg images would help anyone judge raw files anyway?

 

[JimKasson]

Of course, another way to get MTF's way over what you'd expect from a perfect lens on a 100% fill factor sensor is to sharpen the image. And that includes the stealth sharpening performed by some raw developers and most in-camera-JPEGs.

Jim, they tested JPEGs at default sharpening settings.

Then the numbers are pretty useless for judging how sharp raw files would be.

True. But who was claiming jpeg images would help anyone judge raw files anyway?

Tho OP, by the thread title, implied the sensor was being tested. The only way you’d have a prayer of testing the sensor would be to look at raws.

 

[Erik Kaffehr]

https://www.pdnonline.com/gear/cameras/mirrorless/nikon-z7-camera-review/

Hi,

Those tests are more an evaluation of the camera's JPEG processing than an evaluation of the camera and the sensor.

Image Engineering is probably a well respected lab, doing a lot testing for German periodicals and developing gear for camera testing.

But, the test reports discuss results from the cameras JPG engine and that has a lot of value if you shoot JPEG but very little value if you shoot raw.

Best regards

Erik

 

[jfriend00]

https://www.pdnonline.com/gear/cameras/mirrorless/nikon-z7-camera-review/

Hi,

Those tests are more an evaluation of the camera's JPEG processing than an evaluation of the camera and the sensor.

Image Engineering is probably a well respected lab, doing a lot testing for German periodicals and developing gear for camera testing.

But, the test reports discuss results from the cameras JPG engine and that has a lot of value if you shoot JPEG but very little value if you shoot raw.

Well, it's an evaluation of the whole image chain: lens + sensor + readout + JPEG processing + camera settings. One can only draw conclusions about what the whole chain at the chosen settings produces. One cannot draw conclusions about the sensor alone. One can conclude that all the elements of the system must be pretty good to get good overall results, but one cannot conclude anything about how this sensor alone compares to some other sensor.

 

[Rj88]

https://www.pdnonline.com/gear/cameras/mirrorless/nikon-z7-camera-review/

To be honest, it seems the DELTA values are quite a bit off for some colors.

 

[Erik Kaffehr]

https://www.pdnonline.com/gear/cameras/mirrorless/nikon-z7-camera-review/

To be honest, it seems the DELTA values are quite a bit off for some colors.

Hi,

I think that is bad information. You probably have at least a dozen settings on the camera that affect color reproduction.

In general, you don't want a faithful representation of color as long as the dynamic range of your display device does not match the luminance range of the subject. Delta E takes luminance variations into account, so it is a very bad measure for accurate reproduction of color.

In general, saturation is enhanced in raw conversions. Cutting back on saturation would often improve DeltaE, but it may be that an image that is oversaturated may be perceived better than a lower saturation image that is more sensitometrically correct.

Best regards

Erik