dclark
Online
• • • • •
Upload & Sell: On
|
 p.1 #9 · So, is the Z7 II the better camera then? | |
DanielJStein wrote:
Like another photographer said, you are referring to the process of downsampling. While it is true that you can downsample the size of the Z7 files to equal that of the Z6 files, it does not paint a full picture of what is going on behind the scenes.
Natively, the Z7 has a sensor with 4.5 microns of pixel pitch while the Z6 has 6. This means that each pixel on the Z6 is larger and therefore can collect more photons per pixel. Think of it like a bucket. The bigger of a bucket you have, the more things you can put in it. If you have two rooms with a bunch of buckets taking up the same amount of surface area in each, but in one room the buckets are bigger, while the other room the buckets are smaller and there are more of them, which buckets are going to fill up more? The room with the buckets that are bigger. Therefore there is more in each bigger bucket. (More signal, less noise).
This equates to the Z6 being the better low light performer in that regard. In practice, the 6 is giving you about a 1.3 stop "noise" advantage over the 7. I have tested this with both my 6 and 7 and found it to be largely true.
In other words, no matter how you slice it, the 6 is still "better" as a low light camera. Not to say the 7 is bad, because it certainly is still an incredible camera, but if you prioritize milky way or other astro work you would still be better off with the 6. You cannot change the pixel pitch of a sensor simply by software downsampling. It is a physical property created by the size of the sensor divided by its resolution.
Since you benefit of having a few cameras what I would do is use your 7 for daytime landscapes, and your 6 for night work. This is what I personally do with my 6 and 7 and I have found it works great. The extra resolution of the 7 is really nice to have when it comes to landscapes, but I would rather sacrifice the resolution for the benefit of better low light for astro. ...Show more →
---------------------------------------------
Steve Spencer wrote:
Simply put the Z6 II will have just a bit better high ISO files, whereas the Z7 II will have more resolution and a bit better low ISO files. It really isn't a matter of one camera being better than the other camera. Each camera is slightly better than the other camera at what it does best.
That said the bucket analogy isn't being applied quite right here because it doesn't have that downsampling step that others have been talking about. It is that downsampling that let's Z7 II files get closer to Z6 II files. To stretch the analogy imagine you have a third room with buckets the same size as the room with the Z6 II buckets and they you distribute the water from the smaller buckets in the Z7 II room to this new room. The two rooms (the original Z6 II room and the new room with Z6 II sized buckets and water distributed from the smaller Z7 II buckets) will have petty similar buckets and amount of water in the buckets. That is what downsampling does. It redistributes the water from smaller buckets into bigger buckets.
These things can be measured and here is one measurement of the differences in dynamic range:
https://www.photonstophotos.net/Charts/PDR.htm#Nikon%20Z%206II,Nikon%20Z%207II
You will see that at base ISO (64 for the Z7 II and 100 for the Z6 II), the Z7 II has four tenths of a stop better dynamic range. This reflects the better low ISO performance of the Z7 II. Conversely at ISO 3200 the Z6 II has four tenths of a stop better dynamic range than the Z7 II (and this even accounts for downsampling). The difference at high ISO if you downsample is nowhere near a stop and a third, but there is still and advantage at high ISO for the Z6 II.
So, again there is no easy answer here about which camera is better. You will get more resolution and more dynamic range with the Z7 II when you are shooting at ISO 64, but at high ISO the Z6 II will give you a bit of advantage even if you downsample. You will either have to get both cameras or decide which type of shooting is more important for you....Show more →
The best measure of sensor performance is image information.
The pixel SNR is interesting but not a good measure of the image information produced by a sensor.
Photographic dynamic range (PDR) is better but still does not measure the image information.
Consider an ideal sensor that is limited only by shot noise (read noise, pattern noise, etc are zero). The SNR of each pixel is approximately sqrt(number of photoelectrons). If each pixel is divided into 4 pixels, the photoelectrons are also divided up among the 4 pixels. If they equally divide each of the smaller pixels now has a worse SNR that is 2X the SNR of the large pixel. Note that the SNR per unit area is unchanged. That means the photographic dynamic range is unchanged. Which one has the most image information? Clearly the one that provides more information about the spatial distribution of the photoelectrons. That is the one that provides more information about higher spatial frequencies in the image. This is true regardless of how small the number of photoelectrons may be (i.e. how high the ISO may be).
That may be true for an ideal sensor, but what about real sensors. What if each pixel contributes some read noise so that when the number of pixels is increased the amount of read noise in the image increases. The noise in the image is not only shot noise. Photographic dynamic range, which is a measure of SNR per unit area, is a better measure of image information capability than single pixel SNR. But if two sensors have the same PDR but one has more pixels, the one with more pixels produces more image information. If the sensor with more pixels has worse PDR, how much worse can it be and still produce more image information? That's a very difficult question. A full discussion of that would go on for a long time and IMO at the end of the long discussion there would be disagreement.
I agree with Steve that the best unambiguous data we have to compare the noise performance of sensors is PDR, but if the higher resolution sensor is slightly worse for PDR it is not convincing. I also look at the input referred read noise and analyze how far into the shadows we need to go to see a real impact on SNR from read noise. What I find completely useless is the "calibrated eyeballs" that look at a bunch of images and conclude which sensor is better. Unfortunately that is the most common form of sensor comparison on the internet.
Dave
Edit: I corrected a major error. I had written 1/sqrt(number of photoelectrons) where I should have written sqrt(number of photoelectrons).
Edited on Aug 16, 2021 at 10:07 PM · View previous versions
|
