[BigMackCam]

Folks, thanks for all the input so far (and for keeping it all constructive) Interesting stuff, and I look forward to reading and digesting more of it.

At this stage, though, I do have a couple of more specific questions:

My understanding is that greater resolution for a given sensor area should resolve greater detail, assuming the lens out-resolves both examples. So, a 24MP sensor should out-resolve a 16MP sensor of the same size, yes? Putting aside SNR, lens performance, AA filter and other factors for a moment, is it then reasonable to assume that full frame and APS-C sensors (or any two sensors of different sizes) with the same pixel density should be capable of resolving the same level of detail? Further to that, if the answer is "yes", is it reasonable to assume that a larger sensor with lower pixel density than a smaller sensor is technically able to resolve less detail?

Apologies if these are dumb questions Please be gentle

---

Last edited by BigMackCam; 08-13-2019 at 02:13 PM.

[GUB]

When I have time to do so (which should hopefully be in the next couple of days), I must run a couple of simple tests. I own a 2015 Pentax K-3 with 24MP APS-C Sony sensor, and a 2015 Sony A7 MkII with 24MP "full frame" sensor. I see that the DxOMark scores for both cameras are similar, with a very small advantage to the A7 MkII... except for "Sports (Low light ISO)", where the A7 MkII, at 2449 ISO, trounces the K-3's 1160 score. In general lower-light shooting, I've not noticed this difference to such a large extent, but I haven't been looking for it either...

I will be interested in the results. Increased dynamic range was one of the things I noticed when going from the K-01 to the K-1. But they have the same pixel pitch so I presume the gain was from improved tech. How are you going to objectively demonstrate what photoptimist suggests is a 1EV gain in DR? The best I can think of is shouldn't the larger DR have a smaller histogram footprint?

[GUB]

Folks, thanks for all the input so far (and for keeping it all constructive) Interesting stuff, and I look forward to reading and digesting more of it.

At this stage, though, I do have a couple of more specific questions:

My understanding is that greater resolution for a given sensor area should resolve greater detail, assuming the lens out-resolves both examples. So, a 24MP sensor should out-resolve a 16MP sensor of the same size, yes? Putting aside SNR, lens performance, AA filter and other factors for a moment, is it then reasonable to assume that full frame and APS-C sensors (or any two sensors of different sizes) with the same pixel density should be capable of resolving the same level of detail? Further to that, if the answer is "yes", is it reasonable to assume that a larger sensor with lower pixel density than a smaller sensor is technically able to resolve less detail?

Apologies if these are dumb questions Please be gentle

But you can't put aside lens performance - without the lens you have no detail. And quite simply a 4/3 camera lens has to be capable of twice the resolution of a full frame camera lens to give an equal result.

[BigMackCam]

I will be interested in the results. Increased dynamic range was one of the things I noticed when going from the K-01 to the K-1. But they have the same pixel pitch so I presume the gain was from improved tech. How are you going to objectively demonstrate what photoptimist suggests is a 1EV gain in DR? The best I can think of is shouldn't the larger DR have a smaller histogram footprint?

Honestly, I'd intended for my initial test to focus primarily on colour and luminance noise at different ISO settings (say, ISO 100, 1600 and 6400). I'd be happy to test DR, but I've no experience in doing so, and would need advice or instruction. That aside, I'm fairly happy to accept DxOMark's figures of 13.4 EV and 13.6 EV for the K-3 and A7 MkII respectively (since I'm certain DxO's testing methods will be more controlled and reliable than my own). It's a small difference in the grand scheme of things.

I'm also interested to test out detail resolution - for example, distant foliage - between the two cameras, as I suspect the K-3's higher pixel density sensor might have better resolving capability than the A7 MkII (see my post above)...

[house]

The web is full of these discussions. One aspect I don't think has been mentioned is that the size of sensor changes the pressures on the optics. At the wider end my eyes tell me that larger sensors/film and their optics produce more airy, spacious and smooth images. Small sensors end up with a crunched up look. I'm guessing this is a lens issue as I can't see any other reason.

[BigMackCam]

But you can't put aside lens performance - without the lens you have no detail.

Absolutely... and this one aspect of many, IMHO, exposes a weakness in any blanket, unqualified claim that greater sensor real-estate results in better image quality. But that's the purpose of the discussion, and I remain open-minded

By putting aside other aspects, I'm trying to isolate one technical defining aspect of image quality based on pixel density between different size sensors.

And quite simply a 4/3 camera lens has to be capable of twice the resolution of a full frame camera lens to give an equal result.

Assuming the same sensor resolution? You see where I'm going with this, I'm sure...

[Wheatfield]

Does greater sensor "real estate" equate to better image quality?

Yes.
This has been shown to be true time and again, though it is possible to skew results in favour of the smaller format by deliberately sabotaging the quality from the larger format. For example, a really good lens on a small format will give better results than a piece of garbage lens with separated elements will give on a larger format.

If a person is going to have an honest discussion regarding this, it has to be a given that the format is the only significant difference in the comparison.

---

Last edited by Wheatfield; 08-13-2019 at 03:01 PM.

[victormeldrew]

The biggest problem with the whole discussion and the idea of testing it, it this (from my earlier post):

...all other factors being equal...

Unfortunately, it is impossible for all other factors to be equal in the the real world. Even if you had two sensors with the same number of pixels, built in exactly the same way, just in different sizes, the knock on effects of that different geometry affect numerous other factors which also have an impact on image quality. Isolating the effect of the sensor while excluding all others is impossible.

[BigMackCam]

Does greater sensor "real estate" equate to better image quality?

Yes.
This has been shown to be true time and again, though it is possible to skew results in favour of the smaller format by deliberately sabotaging the quality from the larger format.

"Deliberately sabotaging"... hmmm... Perhaps I should have given more context in my original post To be clear, I'm talking about real-world photography, here. Folks buying cameras, the lenses available for those cameras, and their desired use cases. Some will buy the latest and best models, some will buy older and/or more basic or economical models. Some will print large... some (most?) won't. These are some reasons why, in my view, blanket / unqualified statements are potentially misleading.

For example, a really good lens on a small format will give better results than a piece of garbage lens with separated elements will give on a larger format.

For sure, albeit the poles in this comparison are pretty extreme, so no surprise as to the conclusion. But when the optical instruments are more evenly matched, what then? The sensor is key, clearly. What if the optical instruments are outstanding on both, but the full frame sensor is an older 12MP model, whilst the APS-C sensor is five years newer and 24MP, with greater resolving capability, dynamic range and high ISO performance? That's unfair, of course - as extreme as your lens example - but I'm using it to emphasise the need to qualify the statement that "greater sensor real estate equates to better image quality". What about two sensors from the same era... Even if technical data suggest the larger sensor is better than the smaller, does it translate to appreciably better image quality? Possibly (probably, perhaps)... but under what scenarios is that benefit realised, and to what extent? Will the "average" user (if there is such a thing) realise those benefits? Will the person asking "will I get better image quality from a K-1 than a KP" realise those benefits?

If a person is going to have an honest discussion regarding this, it has to be a given that the format is the only significant difference in the comparison.

Since I started the thread, that person is me... and I'd hope you know I'm always honest I'm happy to start off with format being the only significant difference in the comparison. But we should recognise that opinions and advice given to others in these (and other) forums regarding larger vs smaller sensor formats is usually related to practical usage - for instance, where folks are deciding which camera and lens to buy... and then, factors other than sensor format really do matter, and blanket statements become riskier. If a person is going to give an honest recommendation on comparative image quality between sensor formats when asked for advice, these factors have to be considered... IMHO.

[BigMackCam]

The biggest problem with the whole discussion and the idea of testing it, it this (from my earlier post):


Unfortunately, it is impossible for all other factors to be equal in the the real world. Even if you had two sensors with the same number of pixels, built in exactly the same way, just in different sizes, the knock on effects of that different geometry affect numerous other factors which also have an impact on image quality. Isolating the effect of the sensor while excluding all others is impossible.

An excellent observation that doesn't undermine other potentially valid arguments, but quite nicely isolates one fundamental aspect that's rather key to any blanket statement, IMHO...

[photoptimist]

Folks, thanks for all the input so far (and for keeping it all constructive) Interesting stuff, and I look forward to reading and digesting more of it.

At this stage, though, I do have a couple of more specific questions:

My understanding is that greater resolution for a given sensor area should resolve greater detail, assuming the lens out-resolves both examples. So, a 24MP sensor should out-resolve a 16MP sensor of the same size, yes? Putting aside SNR, lens performance, AA filter and other factors for a moment, is it then reasonable to assume that full frame and APS-C sensors (or any two sensors of different sizes) with the same pixel density should be capable of resolving the same level of detail? Further to that, if the answer is "yes", is it reasonable to assume that a larger sensor with lower pixel density than a smaller sensor is technically able to resolve less detail?

Apologies if these are dumb questions Please be gentle

Yes, a 24MP sensor (e.g., a K-3) would out-resolve a 16MP sensor (e.g., a K-5) of the same size (APS-C) both at the print level and the pixel peeper level when attempting to create similar images of the same scene.

A large 200 pixel/mm sensor (e.g., a K-1) would out-resolve a small 200 pixel/mm sensor (e.g., a K-5) at the print level but only match its resolution at the pixel peeper level when attempting to create similar images of the same scene.

Putting aside SNR, a 24 MPix FF sensor would match a 24MP APS-C sensor on both print and pixel peeper level when attempting to create similar images of the same scene. But the FF sensor would tend to have a better SNR which means it would out-resolve the APS-C sensor for scenes involving high ISO, shadow recovery, and very low contrast details.


EXCEPTIONS:

1) for extremely small sensors with high pixel counts, diffraction will ruin resolution both at the print and especially at the pixel-peeper level.

2) for Bayer sensors with low pixel counts or low pixel density, aliasing will ruin resolution unless a matched AA filter is added.


P.S: The question is a little ill-defined in terms of HOW resolution is measured (print vs. pixel-peeper) and WHICH lens-camera pairings used for analysis (same lens vs. equivalent FOV lens).

[GUB]

Assuming the same sensor resolution? You see where I'm going with this, I'm sure...

Yes this is probably where we are thinking differently. I don't see pixel pitch vs lens resolving power as a weaker link situation - a high quality lens still improves a low mp image. One of the clubs I go to that doesn't focus on photography still uses a pretty crappy old LCD projector that would be lucky to be 1200x700. You can easily see the difference of the differing qualities of cameras on it yet it would be easy to say the projector was the weakest link in the presentation of the images.

[BigMackCam]

@photoptimist - thank you, this is really helpful, though it raises some additional questions for me (and you've always been patient with my questions ), so here goes...

Yes, a 24MP sensor (e.g., a K-3) would out-resolve a 16MP sensor (e.g., a K-5) of the same size (APS-C) both at the print level and the pixel peeper level when attempting to create similar images of the same scene.

Got it. But... Is there a print size and resolution (DPI) where this begins to have a practical impact at viewing distances relevant to the print size? I know that's a question with huge variables involved, but I hope you can see where I'm coming from. I'm not asking it to confuse the issue, but rather to understand where the greater resolution of the sensor becomes advantageous in real terms...

A large 200 pixel/mm sensor (e.g., a K-1) would out-resolve a small 200 pixel/mm sensor (e.g., a K-5) at the print level but only match its resolution at the pixel peeper level when attempting to create similar images of the same scene.

OK, here I'm confused... or perhaps not... I understand that at pixel-peeping level the resolution would be matched, but the overall size of the image would be greater from the larger sensor. But how does that result in the larger sensor out-resolving the smaller one? Surely it's resolved to the same level, but over a wider field of view?

Putting aside SNR, a 24 MPix FF sensor would match a 24MP APS-C sensor on both print and pixel peeper level when attempting to create similar images of the same scene.

Right, that makes absolute sense to me :D Wait, no... I don't understand... Larger pixels on the full frame sensor would resolve to the same level as smaller pixels on the smaller sensor? I'm confused (again )...

But the FF sensor would tend to have a better SNR which means it would out-resolve the APS-C sensor for scenes involving high ISO, shadow recovery, and very low contrast details.

Ah, OK... and perhaps this answers my earlier confusion...

[photoptimist]

@photoptimist - thank you, this is really helpful, though it raises some additional questions for me (and you've always been patient with my questions ), so here goes...

You're welcome!

Got it. But... Is there a print size and resolution (DPI) where this begins to have a practical impact at viewing distances relevant to the print size? I know that's a question with huge variables involved, but I hope you can see where I'm coming from. I'm not asking it to confuse the issue, but rather to understand where the greater resolution of the sensor becomes advantageous in real terms...

Tricky! The answer depends on the scene (detailed & high contrast scenes benefit from higher PPI), the lens&camera (sharp lenses & cameras benefit from higher PPI), the post processing (intensive sharpening benefits from higher PPI ), the printer (high-quality printers show better high-PPI results), and the viewer(younger eyes benefit from higher PPI)! Normhead has pretty strong opinions about this -- IIRC, he tends to say that about 150 PPI is all you need and PPI DPI is invisible & over-kill. However, for monochrome text (high-contrast & high-detail), I notice that PPI DPI is better than 300.

Printed at 32"x48", a K-5 image is 100 PPI and a K-1 image is 150 PPI which should be noticeable. Printed at 16"x24", a K-5 image is 200 PPI and a K-1 image is 300 PPI which may be noticeable for some scenes shot with good lenses. Printed at 8"x12", a K-5 image is 400 PPI and a K-1 image is 600 PPI which isn't likely to be noticeable except for images with the sharpest high-contrast details.

OK, here I'm confused... or perhaps not... I understand that at pixel-peeping level the resolution would be matched, but the overall size of the image would be greater from the larger sensor. But how does that result in the larger sensor out-resolving the smaller one? Surely it's resolved to the same level, but over a wider field of view?

This is where we get into arguments about apples versus oranges and whether you have to use the same lens (but different FOVs) to get apples-to-apples or have to use the same FOV (but different lenses) to get apples-to-apples. Are you testing overall resolution with the same lens or testing resolution with the same scene?
And it depends on whether you measure resolution in line-pairs per mm on the sensor or line-pairs per frame-width.

Right, that makes absolute sense to me :D Wait, no... I don't understand... Larger pixels on the full frame sensor would resolve to the same level as smaller pixels on the smaller sensor? I'm confused (again )...

The confusion might depend on whether you are talking about pixel peeping versus print viewing and whether you used the same lens or equivalent FOV lenses on the two cameras.

Ah, OK... and perhaps this answers my earlier confusion...

Hope so!

[Lowell Goudge]

So let's say that your 16x24mm sensor and your 24x36mm sensor are both 24 megapixels, and that with both sensors you are shooting with a lens that can resolve more detail than the sensor. And let's say that both sensors have the same signal-to-noise ratio. Since the 24 megapixels of each sensor has to be downscaled to be viewed on a 1080p or even a 4K monitor, in what sense can you be said to be enlarging the two images by different amounts?

In fact, even if the APS-C sensor had 16 megapixels and the "full frame" sensor had 32 megapixels, in what sense what would you be enlarging by different amounts when you're actually downscaling both to viewing resolution? Surely it's only when you start upscaling that you can meaningfully be said to be enlarging?

I've been sitting here scratching my head while I've been eating lunch, convinced that there must be something obvious here that I'm missing, but I can't figure out what it is. I just hope I haven't ended up with dandruff in my soup.

The issue regardless of film or digital is that you are enlarging both, but until the enlargement exceeds the display resolution, such that you can individually inspect each pixel, on your monitor, or each grain of silver halide on the film , perhaps my standing a foot away from a slide projected on a 4 foot screen You are looking at the composite image of multiple pixels or grains, the consistency between adjacent grains / pixels is what is evaluated as quality