Sensor Pixel Density

After looking at some photos from a D500 shooter, I began to think about the sensor pixel density of the camera. This lead to the question of how it compares to the Nikon cameras I own, have owned and would like to own.

In calculating the pixel densities, I assume that DX sensors are all 16mm by 24mm and FX sensors are 24mm by 36mm. I also assume that the total number of pixels on the sensor is as claimed by the manufacturer.

Here are a few of the figures I came up with:

D700 FX 12.1MP 14029 pixels/sq mm

D750 FX 24.3MP 28125 pixels/sq mm

D810 FX 36.31MP 420149 pixels/sq mm

D500 DX 20.9MP 54427 pixels/sq mm

D7200 DX 24.2MP 63021 pixels/sq mm

Those numbers are quite impressive, IMO. It also shows part of the reason why photos from some of the DX lineup look so good. Besides using the best portion of the image resolved by the lens, it has advanced imaging and control software and a whole lot more pixels to resolve the details of the image.

It makes it even harder not to skip the D500 and save for a longer lens...

WM

WM

There is a lot I do not understand about Pixel density. But as I have been trying to understand it with my old brain. Larger pixels show less grain so more pixels is not necessarily better in all aspects. But more pixels is better for some things. Looks like there is probably a sweet spot between grain and resolution? from my limited experience with just the D7100 and D750 both of which are 24mp I am hoping that 24mp is that sweet spot.

Now the Nikon D4 only has 16.2 MP and look what fantastic images it produces.

Noise/grain only really comes into play when you need to amplify the light (i.e. increase ISO). The greater the pixel density the higher the likelihood that noise is generated - all other things being equal. But that's the catch - things usually aren't equal. As tech advances pixel density can increase while noise decreases. My 16MP D7000 was usable up to about ISO 1600, but the 24MP D7100 was usable at 3200 and even 6400 for me.

BackdoorHippie said:

Noise/grain only really comes into play when you need to amplify the light (i.e. increase ISO). The greater the pixel density the higher the likelihood that noise is generated - all other things being equal. But that's the catch - things usually aren't equal. As tech advances pixel density can increase while noise decreases. My 16MP D7000 was usable up to about ISO 1600, but the 24MP D7100 was usable at 3200 and even 6400 for me.

Click to expand...

Jake, My experience has been that it is far better to increase the ISO and properly expose the image, as opposed to leaving the ISO lower and using post processing to improve an underexposed image. While there are limits to what ISO may be satisfactorily usable for each camera/camera model, I rarely run out of ISO range where I don't have a backup plan for better exposure.

What I'd like to see would be a comparison of a particular sensor, coupled with the software delivered with different cameras, i.e., say the D500 with EXPEED 3 and 4 processing. As well as the earlier cameras with the EXPEED 5 processing... I don't ever see that happening, though.

WM

Don Kuykendall said:

There is a lot I do not understand about Pixel density. But as I have been trying to understand it with my old brain. Larger pixels show less grain so more pixels is not necessarily better in all aspects. But more pixels is better for some things. Looks like there is probably a sweet spot between grain and resolution? from my limited experience with just the D7100 and D750 both of which are 24mp I am hoping that 24mp is that sweet spot.

Now the Nikon D4 only has 16.2 MP and look what fantastic images it produces.

Click to expand...

Don,

For the longest time, I thought the sweet spot was about 16 - 18 MP, since Nikon's flagship cameras were equipped with sensors in this pixel number range. I now believe that we will see that number increase due to camera and camera software improvements. Time will tell, and of course, it also depends upon whether the sensor is either DX or FX sensor equipped.

These are exciting times, but I hope that the camera manufacturers can remain in business to effect these improvements. After all, very few are producing film any more.

I'm interested to hear what other people think about this, especially those like Jake, who really put a lot of thought and effort into this side of photography.

WM

Some time ago I did a comparison based on a some cameras, these included the following camerabodies:
- D610 (FX sensor, 24MP, EXPEED 3 processor),
- D5300 (DX sensor, 24MP, EXPEED 4 processor),
- D5100 (DX sensor, 16MP, EXPEED 2 processor),
- Nikon 1 J5 (CX sensor, 21MP, EXPEED 5A processor)

All captures have been taken with the Micro Nikkor 60/2.8, the camerabodies being installed on a tripod, which was kept in the same place (i.e. Having an almost identical distance between the lens front and the balls)

As regards overall lightconditions, these captures were taken in the late evening (dark outside) with only few light shedding in through a window. The only - and dominant - lightsource was a single led as shown below, but whose lens focussed most of the light to the (white) ceiling.



Before starting with the low light captures, here are some taken with a flash (ringflash mounted on the lens) - this may also give some extra understanding of the crop factor resulting from the different sensors (FX, DX, CX)

D610 - full frame sensor so a crop factor of 1 (i.e. No crop)


D5300 - DX sensor with a crop factor of 1.5, giving the 60mm the same angle of view than a 90mm on a full frame sensor


D5100 - DX sensor with a crop factor of 1.5, giving the 60mm the same angle of view than a 90mm on a full frame sensor


Nikon 1 J5 - CX sensor with a crop factor of 2.7, giving the 60mm the same angle of view than a 162mm on a full frame sensor



Stepping up the ISO started at ISO 100 and was doubled for each extra step (i.e. 100, 200, 400, 800, 1600, 3200, 6400, 12800, 25600).

Captures have been taken with a 1s exposure and a 5.6 aperture (I didn't want to play with all the benefits of the 2.8 aperture, that the lens has, both to get into high ISO, as well as having some extra DOF). With the given lightconditions and these settings it was only starting at ISO800 that one can identify some structures on the non processed captures. So the first series shown here is at ISO1600.

ISO 1600
D610


D5300


D5100


Nikon 1 J5



ISO 3200
D610


D5300


D5100


N1J5



ISO 6400
D610


D5300


D5100


N1J5



ISO 12800
D610 (Hi1)


D5300


D5100 (Hi1)


N1J5



ISO 25600
D610 (Hi2)


D5300 (Hi1)


D5100 (Hi2)


N1J5
(ISO value not available on this camerabody)


All captures have been taken in RAW. No other processing than adding the watermark and converting to JPG was done.

Checking the different captures one can easily see when the different sensors start to add noise or loose DR. Overall, with no surprise, FX performs best as sensors, and so do more recent processors.

What I didn't/couldn't add at that moment in this comparison was the D500. Therefore here some captures taken with the D500 at high ISO:


D500 ISO 3600


D500 ISO 51200

Whiskeyman said:

Jake, My experience has been that it is far better to increase the ISO and properly expose the image, as opposed to leaving the ISO lower and using post processing to improve an underexposed image. While there are limits to what ISO may be satisfactorily usable for each camera/camera model, I rarely run out of ISO range where I don't have a backup plan for better exposure.

What I'd like to see would be a comparison of a particular sensor, coupled with the software delivered with different cameras, i.e., say the D500 with EXPEED 3 and 4 processing. As well as the earlier cameras with the EXPEED 5 processing... I don't ever see that happening, though.

WM

Click to expand...

When shooting raw you're effectively working with all the available light information, so whether you're amplifying it in camera or in post the light you have is the light you have. The question is, "At what point does the camera's introduction of noise do more to distort than to reveal the image?" Up to a certain ISO the properly exposed image will be far easier to work with in post, but when you come out of camera with more noise to clean up than details then I have to believe that you're going to be able to do more with the underexposed image than with the properly exposed but noisier version. It's going to happen at the extreme end of things which is why I work to determine for each of my cameras at what point I'm trading feather details for speckles.

This point is not hard and fast for each body, it only points to where I'm comfortable going for when I shoot little flying critters that I may need to crop down on as well. The fewer the details in the photo the higher you can boost the ISO and come away with something because recovering the details there is far easier. But again, a properly exposed image that doesn't need to be further amplified in post is likely going to win every time in most normal circumstances. I did play with stuff like this once and recall (on the D750) that there was virtually no difference in noise introduced to an image shot at a particular ISO and the same image shot at a lower ISO with the same shutter speed and aperture but underexposed. In other words, a shot at ISO 100 that's 3 stops undereposed and the same one shot at ISO 800 and perfectly exposed look just about identical when I boost the exposure +3 in Camera Raw/Lightroom.

BackdoorHippie said:

... I did play with stuff like this once and recall (on the D750) that there was virtually no difference in noise introduced to an image shot at a particular ISO and the same image shot at a lower ISO with the same shutter speed and aperture but underexposed. In other words, a shot at ISO 100 that's 3 stops undereposed and the same one shot at ISO 800 and perfectly exposed look just about identical when I boost the exposure +3 in Camera Raw/Lightroom.

Click to expand...

Very interesting, Jake. I haven't seen this because I haven't looked for it. Now that you've brought it up, I'm going to give it a go myself.

Thanks for the response; your input is always very respected here!

WM