p.1 #1 · Z8/Z9 with BSI sensor and fewer megapixels?
Do you think Nikon would ever consider a camera that uses a BSI sensor with a lower megapixel count, but with a similar form factor to the Z8/Z9 and with autofocus that works as well as the Z8/Z9 in terms of tracking moving subjects?
Would something like a 24MP BSI sensor provide for better high ISO performance? Would it appeal more to photojournalists and sports photographers (if they can keep a high frame rate)?
p.1 #2 · Z8/Z9 with BSI sensor and fewer megapixels?
The upcoming Zf is the closest thing to that on the horizon, and is also probably our first look at what the Z6III will be (in a different body of course).
The two things you will not get is AF tracking as good as a Z8/Z9 and a blackout-free viewfinder as both of those things require a stacked sensor. AF performance is highly correlated with sensor readout speed in mirrorless cameras, particularly for more demanding applications such as BIF. With the Expeed 7 and the Z8/Z9 algorithms we are sure to see huge AF improvements on the non-stacked sensor cameras but there are hardware limitations that prevent a direct transplant of the Z8/Z9 AF performance in a 'lesser' body. Nikon could offer their flagship AF at multiple price points with DSLRs (D500/D850/D5) for example, because the AF was physically disconnected from the sensor.
The benefits of BSI are far greater on smaller sensors (like those in P&S cameras and smartphones) and the benefits on larger sensors have more to do with readout speed than improving ISO performance, but it does help slightly. Pretty well any modern FF camera sensor these days will be BSI already though. The Z6II has a BSI sensor.
There has been no meaningful ISO performance benefit to lower MP cameras for the last decade or so beyond absurdly high ISOs for niche cases (such as surveillance). All else equal, a higher MP image will look as good or most often better when downsized to match a lower MP image for comparison. In the past, lower MP counts were required for smaller file sizes, faster frame rates, deeper buffers, etc. but all of those challenges have been overcome with bodies like the Z8/Z9. Now that there is no disadvantage regarding either performance or file size to shooting at ~45MP, there is no practical reason to put a lower MP sensor in either of those bodies. You could maybe make an argument for cost as lower MP sensors are somewhat cheaper to produce, but the Z8 is already ~$2,500 USD cheaper than its direct competitor. The lower MP sensors will almost certainly be reserved for the lower-end non-stacked bodies such as the Z6III. At the end of the day, Nikon will do whatever they think will sell the most cameras, but I personally don't think we will see a Z8/Z9 body with a lower MP sensor as there is no longer a reason for it.
p.1 #3 · Z8/Z9 with BSI sensor and fewer megapixels?
I hope so. I prefer the 24 MP sensor in the Z6 II to the 45 MP one in the Z8 for indoor available light photography, which I do a lot. 45 MP is also completely excessive for many if not most applications of photography that I am interested in, so it's just an extra pain in post. I find the Z6 II gives amazingly good image quality in indoor photos and the files are so flexible.
Canon and Sony do make stacked sensor 24 MP models for photojournalism and sports. If this is important to you, perhaps you can consider one of those products (A9 II, R3).
The problem is that it is easier to sell cameras with increasing specs like resolution etc. More, more, more is what humans seem to want.
For me the Z f (if the pictures of the camera are real and not fake) has too minimalistic controls which make for great aesthetics but not very practical operation. The cable release (none!), battery, and memory card of the Z fc are different from the mid and upper end Z cameras and Z f is likely to mimic those. I like the mechanical controls as an addition/alternative to exposure settings but I don't want so many of the other controls of the Z8/9 to be missing. I could live with the controls missing but when the lack of battery and accessory compatibility is added it is just not compatible enough with the rest of the system IMO. However, for now I agree the Z f will probably give an indication of what kind of AF performance can be obtained with a non-stacked BSI 24 MP FX sensor with the Expeed 7 processor. But it probably won't be at all the right camera for sports photography. No vertical grip option, too few and the wrong physical controls, the wrong memory cards, etc.
For photojournalism outside of fast action, I would imagine the Z f (if it materializes around the specs and form given in the recent rumors) would work fine.
oguruma wrote:
Do you think Nikon would ever consider a camera that uses a BSI sensor with a lower megapixel count, but with a similar form factor to the Z8/Z9 and with autofocus that works as well as the Z8/Z9 in terms of tracking moving subjects?
Would something like a 24MP BSI sensor provide for better high ISO performance? Would it appeal more to photojournalists and sports photographers (if they can keep a high frame rate)?
p.1 #4 · Z8/Z9 with BSI sensor and fewer megapixels?
I would switch to a body with Z8/Z9-like performance and a 24mp stacked sensor in a heartbeat. I'd switch even if performance was a bit less... The rumored Z<i>f</i>, if it does meet the "leaked" specs, will give us an indication of what Expeed 7 can do with a non-stacked sensor. I'm sure it will have a leg up over Expeed 6, but don't expect miracles.
The Z6/Z6II files are really a joy to work with. I don't find the Z9 files to be that great of an improvement over 24mp in most cases, and they don't allow quite as much flexibility in post, especially at high ISO; however, the blackout-free EVF, AF, and sheer speed of the Z9 (among many other improvements) makes it a far better camera overall.
While my Z9 is off for service at Nikon, I have been using a D100 and D200 with a rag-tag collection of even older lenses. I can definitively say camera technology has improved a bit in the last 20+ years...
p.1 #5 · Z8/Z9 with BSI sensor and fewer megapixels?
CanadaMark wrote:
There has been no meaningful ISO performance benefit to lower MP cameras for the last decade or so beyond absurdly high ISOs for niche cases (such as surveillance). All else equal, a higher MP image will look as good or most often better when downsized to match a lower MP image for comparison.
Is that true, though? I compared some photos taken with my R3 to some downsampled photos from my Z9, under pretty much the same lighting conditions (shots taken from my office with just the interior lighting), and the R3's images certainly appeared to have less noise at ISO 3200+. Of course, that's not an Apples to Apples comparison, because the sensors have differences other than just the megapixels.
I think it's been hotly debated before, but what's the science between the correlation (or lack thereof) between high iso/less noise with more versus fewer megapixels? Is there any truth to the "fewer pixels means bigger pixels and therefore more light?" talking point?
p.1 #6 · Z8/Z9 with BSI sensor and fewer megapixels?
ahinesdesign wrote:
I would switch to a body with Z8/Z9-like performance and a 24mp stacked sensor in a heartbeat. I'd switch even if performance was a bit less... The rumored Z<i>f</i>, if it does meet the "leaked" specs, will give us an indication of what Expeed 7 can do with a non-stacked sensor. I'm sure it will have a leg up over Expeed 6, but don't expect miracles.
The Z6/Z6II files are really a joy to work with. I don't find the Z9 files to be that great of an improvement over 24mp in most cases, and they don't allow quite as much flexibility in post, especially at high ISO; however, the blackout-free EVF, AF, and sheer speed of the Z9 (among many other improvements) makes it a far better camera overall.
While my Z9 is off for service at Nikon, I have been using a D100 and D200 with a rag-tag collection of even older lenses. I can definitively say camera technology has improved a bit in the last 20+ years... ...Show more →
Just for yucks the other day I was going though old gear and dug out my old D100 and D50. Charged the batteries up and gave em a go. Good lord have I gotten spoiled. Every time you get a new camera you can feel the improvements, but when you rewind the clock 20 years, well, its a bit jarring.
p.1 #7 · Z8/Z9 with BSI sensor and fewer megapixels?
oguruma wrote:
Is that true, though? I compared some photos taken with my R3 to some downsampled photos from my Z9, under pretty much the same lighting conditions (shots taken from my office with just the interior lighting), and the R3's images certainly appeared to have less noise at ISO 3200+. Of course, that's not an Apples to Apples comparison, because the sensors have differences other than just the megapixels.
I think it's been hotly debated before, but what's the science between the correlation (or lack thereof) between high iso/less noise with more versus fewer megapixels? Is there any truth to the "fewer pixels means bigger pixels and therefore more light?" talking point?...Show more →
Yes it's true. DR and ISO performance hit a wall around ~2012 and since then sensor manufacturers have focused on two main things - 1) Keeping ISO performance roughly the same despite increases in megapixels, and 2) Sensor readout speed. There are still minor differences of course, but they aren't enough to base a purchase decision on outside of extreme niche cases.
At a pixel level, all else being equal, the higher MP images will typically be noisier than the lower MP images, but there is no practical scenario where that really matters with modern sensor technology if you are doing a proper comparison. Objective read noise measurements do favor the R3 (vs Z9) everywhere except from ISO 500-1000, but it's more complicated than that when megapixel count is different. If you are comparing your Z9 and R3, you need to downsample the Z9 image to 24MP while viewing the R3 maxed out a 1:1 pixel level and under exactly the same conditions. Further to this, high MP images have finer noise characteristics and better detail retention which is yet another advantage because noise reduction software can do a better job with it and finer noise also is less noticeable in something like a print.
Once you get into the stupid high ISOs like 50,000+, everything looks pretty bad, the lower MP sensors have noticeably less noise even at normalized image sizes, and the higher MP images still retain more detail - so pick your poison. At those ISO levels and beyond though, the usage cases are mostly for things like surveillance where simply being able to identify what is in the image is sufficient (i.e. read a license plate in the dark). Within the ISO range that the overwhelming majority of photographers shoot in, higher MP sensors almost always have the advantage (or are at least not worse) all else equal.
Here is a controlled studio sample at ISOs 3200, 6400 and 12800 in very poor light. I chose a D750, D5, Z9, and R3 to span roughly 10 years and a ~2X Megapixel difference. Note that not only do all 4 sensors look very similar in terms of noise, but the significant sharpness advantage the Z9 images have in terms of fine detail retention, courtesy of the downsampling (same principle as oversampling video footage for maximum quality). To my eye the Z9 actually looks like it retains the best combination of sharpness and noise even at ISO 12,800 - the D5 maybe has just a hair less noise but is significantly softer. The Z9 is clearly better than even the D750 which was a fan-favorite for high ISO performance for a very long time:
p.1 #8 · Z8/Z9 with BSI sensor and fewer megapixels?
Z6III will almost certainly use a BSI sensor but it’s not clear what Nikon’s strategy is in terms of MP count for the Z6/7 III models.
Will we stay around 24MP but with a much faster sensor (though non-stacked) or go higher MP but moderately improved read speeds.
The largest determinants of IQ performance today is the sensor size rather than pixel count as others have suggested. Modern sensors QE are around or above 50% already and 100% represents a perfectly efficient recording of signal so we’re within 1 stop from this theoretical maximum. So a few % differences in QE doesn’t translate to too much visual differences.
But optimisation and read speeds can play a big role since it’s the other side of the SNR equation (the noise part).
We can’t do anything about quantum shot noise so there’s only the read (or electronic) noise part.
Sensors can be optimised for high ISO read but will come at an expense at the other end. Dual gain does help though.
Generally speaking the faster the read, the more noise there is. So this is where there may be some differences as reading many pixels very fast will come with some noise penalties. So a lower pixel count sensor may have less read noise by virtue of the sensor not needing the same bandwidth as a higher pixel count sensor to achieve similar full read speeds.
Bear in mind mirrorless cameras are relying completely on the sensor for all its data from EVF streams to AF calculations to subject recognition, metering and the actual full sensor read for the shot of course. So fast sensors enables a lot of the performance improvements so I think much of the sensor improvements these days have been focused on faster sensors whilst maintains the same read noise performance.
As to AF and subject recognition performance, the camera only needs a smaller subset of video stream for this purpose (within the capabilities of non-stacked sensors) so I dare say the addition of EXPEED 7 alone will bring AF and subject recognition performance close to a Z8/9. But what you won’t get without the stacked sensor are things like blackout free EVF, video performance and omission of the mechanical shutter.
p.1 #9 · Z8/Z9 with BSI sensor and fewer megapixels?
The Z6 II is about 1 stop ahead in dynamic range (dxomark) across equal ISOs compared to the Z9. This reflects shadow noise, so you get equivalent noise in the shadows per equal area of the frame at 1 stop faster ISO setting. It's also ahead in other parameters such as mid-gray SNR (half a stop in mid to high ISO), number of tones distinguishable from noise (about half a stop at mid to high ISOs) and at high ISO settings also color sensitivity (numbers of colors distinguishable from noise) while at base ISO the Z9 is ahead in color sensitivity.
These reflect situations where resolution of the image of neither camera is sufficient to see all the detail that could be seen at the intended display size (which is typical of most applications, I would say).
Comparing with 9-year-old sensors (D750) and cameras which have been equipped with an inferior lens (D750, D5) doesn't make a convincing case since a fair comparison for this purpose is with the closest thing that we have to a future sensor with lower megapixel count and it would be equipped with Z-mount lens. The Z6 II makes a much better case for this purpose as at least it can be assumed to use the same lens in the studio comparison than the Z9.
What's really relevant also is not to only compare using daylight but also use a light source which is typical for low light photojournalistic and sports venues. Some artificial lights have a low color temperature and to obtain normal skin colors, one needs to significantly boost the blue channel which results in increased noise and tends to be something the high megapixel sensors in Nikon's line struggle with. Switch to the incandescent light setting in the studio comparison tool to see how things change. Also, in practice lenses have vignetting and this is emphasized in low-light situations with fast lenses, and to correct for this tends to further emphasize the differences in blue channel noise. All these factors tend to come together in difficult indoor situations which photographers who photograph events frequently run into.
Daylight comparison between Z6 II and Z9 at ISO 25600 (I only selected two cameras, set for comparison at equal display scale):
Incandescent light version (because the left bottom color wheel was underexposed I moved the crop to the top right wheel, so it's not exactly comparable with the daylight case above):
Unfortunately to show what e.g. 2-3 stops of vignetting correction will do to the images I would need to do test shoots myself but I don't have a Z6 II any more, so it'll have to wait for Nikon's next good high-ISO camera. I realized quickly that selling my Z6 II was a mistake, the reason I traded it in was because of the EVF and shutter lags which were driving me nuts but I already miss the image quality of the Z6 II; in indoor available light situations it really shone. I find myself in difficulties with the Z8 in situations which the Z6 II handled with ease. In particular I don't like the color of underexposed areas in the high megapixel Nikons and sometimes when the scene lighting contrast range is high but I cannot use flash for reasons of desiring to be as unobtrusive as I can, I found myself delighted with how easy the Z6 II files were to handle compared to the Z8. The kind of situations might arise when there is strong backlighting from a window, another is when there is a dark church with a chandelier as the main light and I need to include the light source in the frame along with the people. A decent result can be obtained by going with black and white rendering (and heavy local adjustments) but some want color and so I'm stuck deciding whether to shoot e.g. 20% with flash or just to try to use available light, bracket heavily (at high fps) and then do exposure blending to give the shots as much of a chance to be decent. I'll experiment with these in my next events where I need to deal with tricky available light.
And no - these aren't fringe cases, they are quite common.
