Why smaller sensors beat full-frame sensors for wildlife photography

[freixas]

In the real world a typical two camera setup for top wildlife shooters today is the Nikon D850 and the Nikon D500, And while the Nikon D500 is a good camera, to me I personally think that the 850 shots tend to look better

A few years ago several members here compared Their D800's which is 36MP to their D 7200 . As I recall, in all cases the cropped-down D800
files typically looked just a little bit better than the D 7200 photos.

On the Nikon side, the high-end cameras for over 10 years have offered smaller than full sensor captures. For example in 2005 I bought the Nikon D2X which offered a 12 megapixel 1.5 DX sensor,as well as a 2.0 8.2 frames per second high-speed crop mode. In 24 x 36 mm or FX format cameras, Nikon has long offered would it calls the DX crop mode

As usual, thanks for some useful comments, Derrel.

As I'm not familiar with Nikon cameras, I don't know their sensor sizes just from the names. I looked that up, but there is also a qualitative difference in the electronics. I don't have a sense of what the technology differences might be. There's been a lot of great development in sensors recently, especially at the high end (D850, I think, falls in this category). My entire analysis is based on comparing equal electronics.

In the Canon world, the comparsion between the 80D (3-years old tech) and the brand-spanking new DX1 Mark II might be interesting. The 80D gives you a lot more resolution, so it would be interesting to see if the more advanced electronics of the DX1 Mark II would beat it. Two ways to compare: 1) crop the DX1 image to APS-C size and 2) resize the 80D image to 0.56x to match the pixel pitch of the DX1. I'd love to see both.

In looking up some of this, I ran into an article covering the same ground I did: Pixel Pitch on Canon DSLRs - Every Other Shot. The author even makes the same point at the end that pixel pitch comparisons are useful only if all other elements are equal.

 

[freixas]

It will project the same exact SIZE (all thing being equal) but the distortion will vary between the two and the image itself will be far different.

Again, 100% wrong. The distortion will be exactly the same. You might want to draw yourself a picture to try to find where this magic distortion difference occurs. Good luck!

 

[freixas]

Sometimes theory and practice diverge.

Yep. A reality check is always good. If I had a full-frame camera, I'd give it a try.

 

[Derrel]

One thing _I did not see_ ( it might be there, had a slight case of TLDNR) is the presence or absence of an anti-aliasing filter; looking at pixel pitch is one thing..if we add 8 to 12 percent in resolving power for a sensor minus an AA filter, I would think that could tip the scales a bit.

 

[freixas]

One thing _I did not see_ ( it might be there, had a slight case of TLDNR) is the presence or absence of an anti-aliasing filter; looking at pixel pitch is one thing..if we add 8 to 12 percent in resolving power for a sensor minus an AA filter, I would think that could tip the scales a bit.

You didn't miss anything and I appreciate your comment. Something else to consider

Something else I left out is that pixels are not perfectly square and edge-on adjacent to each other. One half the pixel pitch doesn't might not mean four pixels with the same area as one bigger pixel.

I found an interesting article at The effect of pixel size on noise. The author is more informed than me, of course. His conclusion, though, is that shot noise (photon noise) is the same for one large pixel vs. four smaller pixels, as I theorized. Read noise (what I labeled "electronic noise") is another story; it's worse for the smaller pixels, but the author claims that the difference can be small. He provides some comparison photos to support his claims. He compares the Nikon D850 to the Sony A7S. This is almost exactly a 2x pixel pitch difference (the Nikon has the smaller pixels). He points out that, even downscaled to the same size as the A7S, the Nikon images are more detailed (and you can review the results yourself).

 

[RVT1K]

None of this matters if you don't know what you're doing.

 

[RVT1K]

My suspicion is that, at worst, once the data from the smaller pixels are scaled to match the larger pixels, the total noise would be no worse. Feel free to pipe in if you actually know how this works. Alternatively, one could try to determine the difference empirically (shoot the same scene with two cameras; be sure to use the same lens, aperture, exposure, lighting, etc.)..

There is an inherent level of noise in any electronic system, its called the noise floor.
Smaller signals will be closer to this noise floor and have a lower signal/noise ration because of this. Any amplification stage also adds noise.

So a small signal is noisier and boosting it makes it worse.

 

[freixas]

My suspicion is that, at worst, once the data from the smaller pixels are scaled to match the larger pixels, the total noise would be no worse. Feel free to pipe in if you actually know how this works. Alternatively, one could try to determine the difference empirically (shoot the same scene with two cameras; be sure to use the same lens, aperture, exposure, lighting, etc.)..

There is an inherent level of noise in any electronic system, its called the noise floor.
Smaller signals will be closer to this noise floor and have a lower signal/noise ration because of this. Any amplification stage also adds noise.

So a small signal is noisier and boosting it makes it worse.

You might benefit from reading this article I found: The effect of pixel size on noise. It includes some real world comparisons. The author shows his calculations of the effects of both upstream and downstream noise with respect to pixel size. There is also some interesting debate in the comments section there.

 

[RVT1K]

Noise in an electrical system is noise weather it be in the circuitry of a camera, a radio, a piece of test equipment, or what ever. Smaller signals are closer to the noise floor of any device.

And maybe its just the way the author chose his words but in my world all noise is random and will not "cancel each other out completely" when combined.

But I also think a lot of this is a situation where just because you can measure something doesn't mean you can tell the difference.

We encounter this with the equipment my company makes. We can and regularly do make measurement where we can see "a problem" that has zero effect on what the customer is doing.

 

[Ysarex]

So the amount of distortion is minimal in comparison overall, but it is still there. 1.5 to 1.6 for FF v. APS

There is no (zero) difference. Again, the lens has no idea what size sensor it will project on. Format size only matters when you can't fit the entire image of interest (a bird) onto the sensor. A lens that projects a bird as 3 mm x 3 mm on the sensor surface will project the bird identically on sensors of any size. Unless the sensor is smaller than 3mm x 3mm, the sensor size matters not at all. Any distortion caused by the lens will be identical within this 3mm x 3mm area.

Again, wrong.

The lens is a lens is a lens.
That is quite true.
The FORMAT SIZE is what matters in the argument,
But as I said before, the format size is irrelevant if the photographer knows what they are doing.


It will project the same exact SIZE (all thing being equal) but the distortion will vary between the two and the image itself will be far different.

I think this disagreement is effectively solved with this sentence; "The FORMAT SIZE is what matters in the argument, But as I said before, the format size is irrelevant if the photographer knows what they are doing."

The format size is what matters and it is irrelevant.

Joe

 

[freixas]

But I also think a lot of this is a situation where just because you can measure something doesn't mean you can tell the difference.

We encounter this with the equipment my company makes. We can and regularly do make measurement where we can see "a problem" that has zero effect on what the customer is doing.

Actually, I agree. I would welcome real-word comparisons. The link I gave includes some. Sadly, the article dates from 2015--I'd like to see a comparison from some more recent cameras.

Keep in mind that my post is regarding wildlife photography (although, honestly, the wildlife I am thinking of is birds). So if I have a thrush 100 feet away and am shooting with a 400mm lens, what I want to know is if I would be better off shooting with a camera with a smaller pixel pitch. I want to know two things: 1) better off without scaling the image and 2) better off if the larger image is scaled to match the smaller.

My post title is a misleading except in the sense that smaller sensors tend to have smaller pixel pitches (not always, of course).

 

[freixas]

The format size is what matters and it is irrelevant.

Hi, Joe,

I have no idea if I agree or disagree with you. What I would say instead is "The format size does not matter and is (therefore) irrelevant." When I have a shot of a bird that only fills 3mm x 3mm of a sensor's surface, every pixel outside this zone is irrelevant. I don't care if the sensor size is 3mm x 3mm or 8" x 10".

However, my disagreement with Soocom1 was whether this 3mm x 3mm image of a bird is different when it falls on a small sensor vs. a large sensor. The answer is no: the image is exactly the same as it is entirely determined by the lens and the position (not size) of the sensor. As long as the shape and position of a sensor is the same within this 3mm x 3mm area, the projection/distortion is exactly the same regardless of the size or shape of the rest of the sensor. I am assuming that all sensors, regardless of size, are located the same distance from a lens.

 

[Ysarex]

The format size is what matters and it is irrelevant.

Hi, Joe,

I have no idea if I agree or disagree with you. What I would say instead is "The format size does not matter and is (therefore) irrelevant." When I have a shot of a bird that only fills 3mm x 3mm of a sensor's surface, every pixel outside this zone is irrelevant. I don't care if the sensor size is 3mm x 3mm or 8" x 10".

However, my disagreement with Soocom1 was whether this 3mm x 3mm image of a bird is different when it falls on a small sensor vs. a large sensor. The answer is no: the image is exactly the same as it is entirely determined by the lens and the position (not size) of the sensor. As long as the shape and position of a sensor is the same within this 3mm x 3mm area, the projection/distortion is exactly the same regardless of the size or shape of the rest of the sensor. I am assuming that all sensors, regardless of size, are located the same distance from a lens.

I was just noting the problem with logical contradiction there.

I'm not entirely sure what he's saying. I understand how objects in a photo are skewed (shape distorted) toward the edges of the frame relative to the center and I'm aware of this problem occurring with all camera formats i.e. it's a camera problem we've all lived with since day one. So I'm not sure what he's saying about that problem relative to working between different formats. I do find it odd to bring it up at all in this context because the problem diminishes with increasing lens focal length and of all the various sub disciplines in photo those most likely to use the longest of all focal lengths are wildlife photographers who in that context have every other possible thing to worry more about than that.

Joe

 

[freixas]

The format size is what matters and it is irrelevant.

Hi, Joe,

I have no idea if I agree or disagree with you. What I would say instead is "The format size does not matter and is (therefore) irrelevant." When I have a shot of a bird that only fills 3mm x 3mm of a sensor's surface, every pixel outside this zone is irrelevant. I don't care if the sensor size is 3mm x 3mm or 8" x 10".

However, my disagreement with Soocom1 was whether this 3mm x 3mm image of a bird is different when it falls on a small sensor vs. a large sensor. The answer is no: the image is exactly the same as it is entirely determined by the lens and the position (not size) of the sensor. As long as the shape and position of a sensor is the same within this 3mm x 3mm area, the projection/distortion is exactly the same regardless of the size or shape of the rest of the sensor. I am assuming that all sensors, regardless of size, are located the same distance from a lens.

I was just noting the problem with logical contradiction there.

I'm not entirely sure what he's saying. I understand how objects in a photo are skewed (shape distorted) toward the edges of the frame relative to the center and I'm aware of this problem occurring with all camera formats i.e. it's a camera problem we've all lived with since day one. So I'm not sure what he's saying about that problem relative to working between different formats. I do find it odd to bring it up at all in this context because the problem diminishes with increasing lens focal length and of all the various sub disciplines in photo those most likely to use the longest of all focal lengths are wildlife photographers who in that context have every other possible thing to worry more about than that.

Joe

Thanks for the clarification.

Again, my point is that it really doesn't matter if I stick a wide-angle lens or a telephoto on a camera. The focal length comment is a red herring. If an image fits on an APS-C sensor, then this same image will fit on a full-frame sensor with exactly the same projection and distortion. Let's say that there's a certain amount of light drop-off near the edges of the APS-C sensor. The same image won't be near the edge of the full-frame sensor, but it will be identical, with the same exact light drop-off in the same places.

One way to picture this would be to imagine that we took the full-frame sensor and applied tape (ick!) to cover up the portions of the sensor larger than the APS-C size. The APS-C sensor and the exposed portion of the FF sensor occupy exactly the same space and shape. The lens neither knows nor cares what it is projecting onto (remember, we use the same lens in both cases). Unless you believe in magic, the image formed on either sensor will be exactly the same.

 

[Derrel]

I want to see actual cameras compared, with real lenses, comparing Canons with Canons, Nikons with Nikons, Pentaxes with Pentaxes. Be aware that a six-inch mis-focus drops a 50 MP camera down to 4 to 6 MP of "resolution". AF system capability in sometimes overlooked...read the reviews of the latest FF Pentax K1 Mark II... it has slow AF for action work.

Theory is fine, but in-field resolving power is affected by more than just "sensor size".