Diversity and Resolution

[xpatUSA]

Threads involving discussion of resolution can get long, heated and often fade away after hundreds of posts - but with no real conclusions ever being drawn. Why do people so rarely agree, especially when comparing this camera with that? Surely there must be some Unified Concept that would help us all understand what should be quite a simple subject?

My own research has ranged far and wide, from learned papers to lunatic rants on [of course other] fora. For a while, I was convinced that the less resolution, the better. Now, I think it could be the more the better!

The problem, IMHO, is that we photographers are diverse and our tools equally so. Even our preferred outputs are different. And, therefore, posts on the subject - so often made without qualification - are diverse. A few examples:

"there's no difference between a D700 and a D800E" - from a person who never prints anything.
"couldn't do with anything less than 48MP" - prints at 24" x 36" and stands a foot away to view their work.
"1MP is all you need" - from he who sells watches on eBay ;-)
"I'm waiting for a 100MP micro-4/3" - from someone who has never heard of diffraction.

Is there a "Unified Concept"? If so, what might it be?

[Bobobird]

There IS a unified "something" - all have cameras.

Now if the manufacturers advertised by the size of print that can be had I'd bet that everyone would go for the largest even though they never print anything.

[Geoff F]

I think it may be beneficial to consider two different aspects here.

The maximum number of pixels on a sensor and the best 'resolution' (pixels per inch) for a number of end uses. For example, uploading to the internet or printing on various paper types.

As cameras get more powerful (total pixel numbers) my computer struggles to keep up and I often ditch a great number of those pixels when resizing for particular end uses.

Then, I find a case where I have to do a considerable crop; as often happens with wildlife shots. Now I'm glad to have those extra pixels.

[xpatUSA]

I think it may be beneficial to consider two different aspects here.

The maximum number of pixels on a sensor and the best 'resolution' (pixels per inch) for a number of end uses. For example, uploading to the internet or printing on various paper types.

Not forgetting the new generation who have never seen paper, and go "Monitor?" like they go "Elvis who?" ;-)

Samsung Galaxy SII 'phone: 219 ppi (800x400px, 4.27" diagonal).
Samsung Nexus 10 thingy: 300 ppi (2560x1600px, 10" diagonal).

[MrB]

Surely the "Unified Concept" is the final image. If that satisfies the photographer, then resolution, along with all the other possible variables, must be OK. And if the final image has an impact on other viewers, then that is a real bonus.

Philip

[RustBeltRaw]

To my mind, the Megapixel Wars have a straightforward solution if you look at users individually. Some people make the aforementioned 24x36in prints with 12in view distances, requiring nothing less than Hasselblad medium-format digital bodies. Some people only post to Facebook, requiring nothing more than a cell phone camera from 5 years ago. However, the photographer with the Hasselblad has a much larger margin for error. He can re-frame, crop, and drastically downsize with impunity. The photographer with the cell phone will not be able to turn their grainy 4MP snapshot into a museum-quality, large-format print (though some of the exhibits I've seen feature photographers who seem to revel in crummy cameras).

I'd say the unified concept is simply where your photos will wind up, how much you need to spend to have the minimum number of pixels required, and how much larger you want to go to increase your margin for error. Granted, this is ignoring large-photosite advantages like a higher SNR, but I think it's the most important factor when considering your camera's resolution. Pixels are not a panacea, but when you need more, you're stuck.

[Inkanyezi]

I think sensor properties is much more than megapixel count, and that technicians working on those chips are not completely out of their minds. When looking back through a number of years, we can see, that not only megapixel count has increased each year, but also a much more important property to image quality, dynamic range. And at the same time as megapixel count has increased, noise has in fact decreased.

And the "noise" argument is clouding the vision of some of us. Noise is largely moot. Just as the silver image was built up of what we used to call "grain", which was not in fact the silver grains, but patterns created by them in the gelatin over the surface of the negative. Now in the digital age, noise is an inherent property of digital images, contributing to our perception of the photos, just as the "grain" was part of the gelatin/silver photograph.

And smaller photosites denser packed do NOT create more noise in the image. That is a completely moot point. Each one of four photosites packed into the same area as one in another less dense sensor will be more noisy, relatively. But when you bundle them together, noise is not adding up linearly, but some of the noise of some photosites will cancel noise from others. So the net result is about the same noise from the same area whether it has 1, 4, 9, 16, 25, 36, 49 or 64 photosites. There is even a phone camera with forty megapixels, and I swear, it is not noisier than a one megapixel cellphone camera. Over the years, sensors have become less noisy, so a µ4/3 sensor of today may display less noise (and more dynamic range) than a "full frame" sensor of ten years ago.

Sensor area is a factor that is more important to image quality than megapixel count. A larger area will permit a longer focal length for any particular angle of view, and a longer focal length lens mostly is sharper than a shorter focal length lens, for various reasons. Diffraction as well is mostly a moot point, even though it might become a bit more important with very small sensor sizes. Again, sensor size and focal length are more important than pixel count.

There are undeniable advantages also with low pixel count sensors, because of faster readout and less storage, but it is not dynamic range or noise levels. Neither is it image sharpness. Low pixel count sensors are more prone to moire than high pixel count sensors, and when the pixel count is very low, anti-aliasing will blur the image considerably. So there definitely is a pixel count that will be a limit for acceptable image quality, depending on the size of the final image as well as viewing distance. My take is that it is within the region of four megapixels divided or multiplied by four. Most computer displays don't show more than two megapixels, and even for prints more than eight is mostly overkill.

So the technicians have thrown in many more pixels than we actually need, and the only disadvantages are slower readout and a requirement of larger storage space, as well as a need for more computer power in PP. But our multi-megapixel sensors are not noisier than sensors with fewer pixels. In fact, for each year, they have become better in two crucial respects, noise and dynamic range. The megapixel race is not over yet, and there is still one thing that may improve sharpness with more megapixels. A few more megapixels and we won't need the AA filter any more.

[rpcrowe]

There IS a unified "something" - all have cameras.

Or some might have smartphones or tablets or

However, bazillion megapixels, gazillion ISO capabilities and articulating LCD's are what sell today's morphodite still/video DSLR cameras...

I just saw a friend's hundred and twenty five dollar Fuji P&S camera that has voice recording ability to identify his shots. Does my, well over one thousand U.S. Dollar, Canon 7D have this capability?

HECK NO! But, you can shoot a frame of RAW whilst capturing in JPEG or shoot a JPEG while capturing in RAW (Whoop Di Doo!).

You can also print directly from the camera with your 7D (How many of you 7D owners have used that capability?)

I have a early 2000 Olympus C5050Z P&S that has dual memory card capability. Does the 7D have this? NO!

[jcuknz]

A camera is a device for recording light patterns ... a cellphone in camera mode does this so it is a camera as well as a broadcasting device in its communication mode.

[drjuice]

The important thing to remember is that each of us may have two or three different reasons for choosing a particular camera, and I can guarantee that the reasons for which I chose my DSLR have not ever been posted on this site by anybody but me. So, from my perspective, I think it's safe to say that I'm probably not interested in arriving at a single common set of reasons.

When I bought my DSLR, I did look at 11 different vendors' DSLRs. I found that I could pretty well configure the same properties for all the pixels, range of lenses, range of ISO, range of shutter speeds, and nearly all features with all the available vendors at that time. So, having determined that the proverbial dime's worth of difference didn't exist, I went ahead with the vendor on whose cameras I could use my existing set of lenses which would wind up my being able to put together a pretty close to complete setup for about $7000 less than if I'd had to lay out the money for the same set of lenses by a different manufacturer. The other detail was that I wanted to be able to use my in-camera memory devices for storing images that I'd already been using. As I said at the outset, neither of these has been identified by anybody else as important.

And, you'll note that I have NOT asked this site for advice on disposing the film gear I have. I'm actually still using both film camera bodies for microscopy.

I'm jes' sayin'....

virginia

[Shadowman]

There's only one way to settle this, a comparable challenge where everyone photographs the same object (well not the exact same object) an egg. Post your example, include all exif data, shooting setup and let's settle this once and for all.

[xpatUSA]

Personally, I am leaning toward the view too that the resolution of the final image is a possible unifying element, as has been suggested already in a post above. Which means that any comparison between cameras should be made with the same output image and the same post-processing.

As an aside, I recently compared a close-up scene, shot both with a Sigma SD10/Kit lens (2268x1512px) and a Panasonic G1 micro 4/3 w/kit lens (4000x3000px). The G1 image at 100% was a bit soft but, after re-sampling 50% to about the same height as the SD10, the G1 took the prize for best acutance. The reasoning as to why that should be: each SD10 image pixel measured the simple average illuminance at the sensel. But the G1 re-sampled image pixel represented Photoshop's much better (than simple averaging) algorithm when combining four (in this case) original pixels into one.

For a long time I advocated Sigma LO res images for best acutance in web images but, more recently, I find that Sigma HI res images down-sized are better.

[Inkanyezi]

There's only one way to settle this, a comparable challenge where everyone photographs the same object (well not the exact same object) an egg. Post your example, include all exif data, shooting setup and let's settle this once and for all.

I'll bite. How would you like your egg?



post edit: Not that I included any EXIF, Raw Therapee doesn't supply that. And I shrunk it to 2 megapixel, not to exaggerate. It doesn't make a lot of difference anyway. It's taken in my kitchen, in the light from the window, OM-D at ISO 200 with a Trinar 50 mm stopped down to f/16 at about ½ second. Not even in the 16 megapixel image will any noise be evident other than at pixel peeping, but there isn't much more detail than in the shrunk image. As long as they are made for web presentation, pixel count doesn't count.

[FrankMi]

I'm not sure that there is a Unified Concept any more than we are cloned copies of each other. We all have different ideas of what makes for the best camera/lens and the manufactures try their best to appeal to our lusts and provide whatever it takes to outsell the competition.

In a way, most of these kind of discussions are like the handyman that wants to buy a drill. Which one is best, with just the right features and at a price he can afford? The truth is that he doesn't really need a drill at all. HE NEEDS A HOLE!

In our case, we can fantasize about this camera/lens and that, and for some folks, that is all they want to consider. For others, the camera/lens is not the 'be all' and 'end all' but rather, like the drill, simply the tool they use to get the result and the result is what really matters.

Perhaps that is why some of the best images are taken with whatever camera is handy and not necessarily the best or most expensive camera on the market.

[xpatUSA]

Don't egg me on . . . ;-)



Panasonic Lumix DMC-G1 (micro-4/3"), 4000x3000px JPEG, 1/50sec, f/11, ISO 100.

Cropped and re-sampled 50%, sharpened 0.3px, 85%, 0 threshold in PSE6.

[DanK]

I don't think there can be a unified answer because we do different things. What would be nice, however, would be a reasoned and informative discussion of tradeoffs for different uses. E.g., I do both night photography and macro, and the ideal for those two, IMHO, is not the same. I would be much better off with a 7D for macro (far more pixels on the subject at minimum working distance) than with a 6D, but the reverse is true for night photography (the 6D has far better noise characteristics). It would also be nice to have a reasoned discussion of the impact of various factors. E.g., in my experience, diffraction does have some noticeable effect at very small apertures. (Even Ansel Adams, who founded the f/64 group, talked about this). However, in most cases--most apertures and print sizes--its effects are trivial or even too small to be noticed. Nonetheless, there are people on the web saying things like "you should never shoot above f/8."

On a similar note, I can't count how many times I have heard people say that crop sensor cameras are inadequate for landscape use. Over the past couple of years, I have had the opportunity to spend a day working with two professional landscape photographers, one widely published. Lo and behold: they were both carrying crop sensor cameras.

[Brev00]

Why do you seek a unifying concept? Einstein tried and failed. What makes you think more mortal humans can do better? In the end, you are your own unifying concept. Efforts to persuade others to your particular point of view are bound to fail. So, buy into a system and be happy.

[xpatUSA]

Why do you seek a unifying concept? Einstein tried and failed. What makes you think more mortal humans can do better? In the end, you are your own unifying concept. Efforts to persuade others to your particular point of view are bound to fail. So, buy into a system and be happy.

Thank you, Larry.

Perhaps I seek that single property that says it all but without a welter of MP, MTF, LPPH, lp/mm, ppi, dpi, CoC, c/p, c/deg and whatever else we use to describe "resolution", rightly or wrongly. It's almost worth a separate thread: "how many ways are there to measure resolution?".

Humans can always do better, IMHO.

Is "buying into a system" really your suggestion or was it said to make a subtle point ;-)

[revi]

Thank you, Larry.

Perhaps I seek that single property that says it all but without a welter of MP, MTF, LPPH, lp/mm, ppi, dpi, CoC, c/p, c/deg and whatever else we use to describe "resolution", rightly or wrongly. It's almost worth a separate thread: "how many ways are there to measure resolution?".

Humans can always do better, IMHO.

Is "buying into a system" really your suggestion or was it said to make a subtle point ;-)

For starters, you are mixing together definitions on how to define resolution (CoC, MTF) with actual units of resolution (ppi, dpi, lp/mm, LPPH)...
And one could wonder about the use of resolution for sensor pixel counts.

Mixing up concepts and definitions usually doesn't lead to useful discussions, let alone useful results.

[xpatUSA]

For starters, you are mixing together definitions on how to define resolution (CoC, MTF) with actual units of resolution (ppi, dpi, lp/mm, LPPH)...

And one could wonder about the use of resolution for sensor pixel counts [indeed one could, or vice-versa for that matter].

Mixing up concepts and definitions usually doesn't lead to useful discussions, let alone useful results.

If, by "mixing up" you mean that I am confused by those units - I most certainly am not, and I deplore most of them as being incorrect. Please note that I said "rightly or wrongly" - which you left out of your quote above, for some reason. Neither am I personally mixing up anything, my OP was a simple question.

In my opinion, the most useful measure of spatial resolution in any image is edge response in units of pixels. Any measure that involves frequency (e.g. lp/mm) is strictly incorrect because it involves a transformation, e.g. the two transforms need to derive MTF from an edge response, ESF -> LSF -> MTF.