[thePiRaTE!!]

Yes it is, it *is* a real advantage in optical resolution... *if* the *lens* is capable of imaging sufficient detail for the finer sensor/film granularity.

Of course in the real world this is not necessarily so, or not even often so, since digicam manufacturers often boost sensor resolution to advertise "MORE MEGAPIXELS!!!" without improving the lens.

With the 1x1mm sensor (having the same megapixel resolution as the larger sensor), you would have extremely small pixels, << 1 micron^2 in area. The lens couldn't project an image with sufficient detail to provide much contrast between adjacent pixels... so in this case you'd be "wasting" the additional area of the lens's image circle. (Also with a 1x1mm sensor you'd have massive noise issues, and difficulties due to diffraction at the pixel microlenses, but that's a whole 'nother can of worms.)

See my previous post for why EITHER the lens OR the sensor/film can ultimately limit the optical resolution of the complete camera system:

Case 1, "Cropping the sensor is optical zoom:" If your lens projects finer detail than your sensor can record, then you'll get more detail by using a cropped sensor with smaller pixels (assuming you can keep noise under control).

Case 2, "Cropping the sensor is like digital zoom:" If your cropped sensor can record finer detail than your lens can project, then you'll get more detail by using a crummier lens with a smaller FOV/longer focal length.

Anyone noticed how Photozone.de features the disclaimer that "lens performance data are only comparable within the same camera system"? That's because it's the combination of lens and sensor that limit the optical capabilities...

Interplay between the sensor and lens resolution aside, my original point is that 200mm is 200mm is 200mm regardless of how big the sensor that is catching the light is. Having the (misleading marketing jargon) "and a focal length equivalent to 300mm in the 35mm format." makes it seem like the smaller sensor is somehow inherently advantageous in some way, that the lens is offering what a 300mm would offer to the guy using 35mm. The reality is that its still just cropped, same as the wide end. Free loss of image off the sides - what a bargain. Nobody jumps up and down because 20mm now looks like 30mm. In fact, it still looks like 20mm, just with less image on the sides. The center isn't magnified, its just the sides are gone. Same for the long end. Verges on false advertising, honestly.

If we work WITH the smaller pixels scenario, can capture greater detail lens willing, and blow up the resulting image to fit the SIZE of the full 35mm frame image, effectively turning a 200 into a 300 via digital zooming, you've still lost out on the light that went bye-bye off the sides of the smaller sensor and you will have achieved equivalent magnification of a longer lens, still at the cost of the FoV of what the 300 would give on the 35mm frame.

It does't matter how you slice it, the guy painting with a 50% larger canvas can make bigger pictures (better or not aside). You can't dodge this bullet.

Ultimately, there is a sweet spot of pixel density, thats still being acheived. But I think conversely, the ultimate size of the sensor in the 35mm format is known. Its a matter of cost at this point. I think the optimum sensor is probably close and ten years from now we won't be discussing mega-pixel, but some index on how they handle dynamic range and noise, etc.

[moxfyre]

Interplay between the sensor and lens resolution aside, my original point is that 200mm is 200mm is 200mm regardless of how big the sensor that is catching the light is. Having the (misleading marketing jargon) "and a focal length equivalent to 300mm in the 35mm format." makes it seem like the smaller sensor is somehow inherently advantageous in some way, that the lens is offering what a 300mm would offer to the guy using 35mm. The reality is that its still just cropped, same as the wide end. Free loss of image off the sides - what a bargain. Nobody jumps up and down because 20mm now looks like 30mm. In fact, it still looks like 20mm, just with less image on the sides. The center isn't magnified, its just the sides are gone. Same for the long end. Verges on false advertising, honestly.

Okay, but what does "focal length" actually MATTER or MEAN? How does affect the image? Here's how: It merely defines the angle-of-view, and that only in relation to a specific size of sensor or film...

And what are you using "magnification" to mean? Technically, magnification refers to the dimension of the image relative to the object being imaged. But you seem to be using it to suggest some level of rendering detail...

I'll say it again 20mm on an APS-C sensor will not look like 20mm on a FF sensor if the APS-C sensor has the same number of pixels. It will capture as much detail as 30mm would on the FF. This is /assuming/ that the lens has sufficient resolving power to capture finer detail than the FF sensor could record.

I'm starting to feel like a broken record here ... A cropped sensor can be like optical zoom OR like digital zoom, depending on whether it's the sensor resolution or lens resolution that is limiting the complete system.

It does't matter how you slice it, the guy painting with a 50% larger canvas can make bigger pictures (better or not aside). You can't dodge this bullet.

Sure he can make bigger pictures. But what if he has to use a 50% larger brush in going to a 50% larger canvas? Then he can make bigger pictures, but not more detailed ones...

That is a better analogy to the case of enlarging/reducing the sensor while preserving the total number of pixels.

Ultimately, there is a sweet spot of pixel density, thats still being acheived. But I think conversely, the ultimate size of the sensor in the 35mm format is known. Its a matter of cost at this point. I think the optimum sensor is probably close and ten years from now we won't be discussing mega-pixel, but some index on how they handle dynamic range and noise, etc.

What do you mean, "the ultimate size of the sensor in the 35mm format is known"? It's known to be 36x24mm .

Do you mean there's some practical limit to how many megapixels we can cram in there? You betcha... it's limited by noise and dynamic range, but more fundamentally by the diffraction limit. The individual pixels can't get smaller than the wavelength of visible light (400-700 nanometers).

I agree, once we reach that limit, megapixels won't be much of a selling point anymore and we'll worry more about DR, noise, energy consumption, etc. Sort of like how megahertz aren't much of a selling point for microprocessors anymore... now it's all about parallelism, energy efficiency, on-die peripherals, etc.

[thePiRaTE!!]

We're all typing at the same time, lol. Ok, I'll show you what I mean graphically, later. We're seeing things from different perspectives, saying the same things over and not getting our points across. Instead of rushing a reply, I'll continue once I return from work... gah...

And yes "What do you mean, "the ultimate size of the sensor in the 35mm format is known"? It's known to be 36x24mm" Thats what I mean.

[aegisphan]

Well, the resolution of image sensors, film or digital, can be quantified using MTF. The resolution of lenses can also be quantified using MTF, though since you have to "put" the image somewhere, it depends on which sensor you use to measure the lens's MTF The imperfect resolution of the lens and that of the sensor/film combine to limit the total resolution/sharpness of the system. It's not always clearly possible in practice to identify one or the other as most limiting... but sometimes it is.

For example: put a lens on the 6mp K100D, and then on the 10mp K10D, and take pictures of the same scene. Does the K10D's image resolve more detail? If yes, then the optical resolution is more limited by the sensor than the lens. If no, then the opposite.

They certainly can be quantified using MTF chart. However, there is no real experiment out there showing that current digital sensor has the same resolution as film does. I brought that up since in your example, you assume both film and the sensor have 5MP resolution.

As for lens, there are some that can out-resolve the current digital sensor, while some others can't. I know that is what Photozone is trying to quantify.

Actually, I thought we were arguing about whether a 200mm lens on an APS-C sensor would produce more/less/same resolution as a 300mm lens on a FF sensor.... ?

We were and still are.

And I explained why any of those is possible, and why the sensor/film size does affect the resolution of the complete optical system. (You're right, the resolution of the lens itself does not change.)

It's not an unnecessary variable. It's utterly crucial.

To me, the lens resolution variable is unnecessary because it's obvious an inferior lens can't out-resolve the superior lens, given their focal length ratio isn't too far apart.

I can't tell you the exact ratio when the inferior, longer focal will have the same resolution as the superior, shorter focal at the same subject-to-sensor distance under the same FOV (crop the shorter focal to the same FOV of the longer one). But we definitely can calculate that for a given pair of lenses.

Of course, I also assume the lens's resolution as the limiting factor. If I don't, it's because of the incapability of the medium, and not the lens. And so my argument would become a moot point.

At the current state, lenses tend to be the limiting factor, not the sensor/film (few might exceed at the center's resolution, but the corners' are still lacking).

However ...

As I've stated, if the lens's optical resolution is limiting, then shrinking the sensor will be like "digital zoom" (narrower FOV, no extra detail gained)... if not, then shrinking the sensor will be like "optical zoom" (narrower FOV with extra detail).

But sure, let's imagine 200mm and 300mm lenses which have equal resolution in terms of MTF50...

Then we agree on that. If lens is the limiting factor, then we would have digital zoom. .

I still don't know what "reach" means...

Leaving aberrations and distortion aside, the ability of a lens to reproduce a scene can be described by:
(a) the vertical and horizontal angles of view subtended by the sensor
(b) the level of sharp detail projected onto the sensor (quantified by MTF)

What does the notion of "reach" include that isn't covered by angles-of-view and MTF???

I think you forgot the subject's magnification. The subject's magnification is not the relative size of the subject projection and sensor size; it's the size of the subject projection. Since we are talking about the projected image on the sensor/film, we have to assume that the registration distance has to be the same.

At the same subject distance, the 300mm has more "reach" or projects a bigger image on the sensor than the 200mm. So if both lenses have the same resolution, the subject is now covered with more lines, and thus has more details.

I guess "reach" is more of a layman term. I used it because some of the materials I read used this term.

And...

Also, you're talking about the resolution of the lens itself. I agree completely... if both lenses have the same MTF (let's call it 50% contrast at 30 lines/mm)... then the 300mm lens will produce more detail of a distant object, over the entirety of its image circle.

We come to an agreement at this point .

Which is why, again, it's important to decide whether it's really the lens or the sensor limiting the system's optical resolution (and it practice it will be a combination ). If all our sensors had infinite resolution, and all our lenses had poor resolution, you would gain nothing by putting a FF lens on a crop sensor... and in fact you would LOSE part of the image circle.

That's what I assume as stated above. I should've made it clearer in my previous post. :ugh:

Yeah, I think we basically agree here 8) Except that you are more skeptical about whether it is really the sensor that is ever limiting the total resolution of the system... Which you may be right about in many cases. Though with a sufficiently good lens (or sufficiently crappy film/sensor!!) the opposite can be true.

This can be tested quantitatively. Get a 200mm FF lens, a FF film camera with appropriate high-quality film, and an APS-C camera.
(1) shoot an MTF test chart with the FF camera, standing at a distance of 10m from the chart, and scan it.
(2) shoot the same test chart with the APS-C camera, standing at a distance of 15m from the chart.
(3) crunch the numbers to determine MTF50.
(4) If the FF and APS-C camera resolve equally fine details (in terms of object dimensions), then the cropped sensor is like optical zoom.
If the FF camera resolves details that are 1.5X smaller than the APS-C camera, then the cropped sensor like digital zoom.
If it's somewhere in between, well then it's somewhere it between

I totally agree with you here.

I'm arguing at a more specific point, while you already covered everything right there .

Looking at all the MTF charts (or similar method), we can see that few lenses can exceed the tested sensor/film (photozone, popphoto, photodo, ...). If so, I think we can safely say that most of the time, the crop FOV would not give the lens the magnification power of a longer focal lens.

[moxfyre]

Looking at all the MTF charts (or similar method), we can see that few lenses can exceed the tested sensor/film (photozone, popphoto, photodo, ...). If so, I think we can safely say that most of the time, the crop FOV would not give the lens the magnification power of a longer focal lens.

Yes, we do agree!!

Most of the SLR sensors today have more resolution than most of the SLR lenses... so yeah, we're closer to digital zoom than to optical zoom. Somewhere in between, at least, since the combination of lens effects and sensor effects is frightfully complicated It may have been more of an issue 8-10 years ago... digital cameras had looow pixel counts, 2-4 MP, which their lenses could easily max out. So shrinking the sensor in all dimensions would actually improve the optical resolution (at the cost of narrower FOV, of course).

I would say that it is unclear how good the optical resolution of today's best lenses are. You have to put them on a camera to actually measure it, so if it's too high, it's bounded by the limitations of the sensor again. *sigh* So, for example, Photozone.de measures the 77 Ltd to have a center MTF of ~2400 LWPH, which would give an effective resolution of 8.6 optical megapixels if it held across the frame. And I believe they use a K10D for their Pentax tests, so they've totally maxed out the sensor's resolution. In real-world measurements, it's considered "excellent" to get half the optical resolution of the sensor's rated # of megapixels. They might well measure a higher MTF50 if they put the 77 Ltd on a K20D body.... ugh, and now we're back where we started.

[Gimbal]

So does this mean that if the sensor has a higher resolution then the lens, then there is really no point in mounting a 2x (for example) teleconverter?

[thePiRaTE!!]

So does this mean that if the sensor has a higher resolution then the lens, then there is really no point in mounting a 2x (for example) teleconverter?

Nothing to do with that. The tele magnifies the image projected to the film/sensor by its multiplication factor. Works just like a magnifiying glass. Extends the reach of your lens

[moxfyre]

Nothing to do with that. The tele magnifies the image projected to the film/sensor by its multiplication factor. Works just like a magnifiying glass. Extends the reach of your lens

On the other hand, if the resolution of the *lens* is maxed out, there's no point in adding the TC... since you're just magnifying the aberrations and not getting additional detail from the scene. This is the reason it's often advised not to use TCs with zoom lenses, but only with primes.

And by "extends the reach", you mean "narrows the FOV". PuhLEEZ

[thePiRaTE!!]

Of course it narrows the FoV. So does using a 400 instead of a 200, or a 200 x2 via TC.

I define FoV as the width of the your field of view expressed as an angle. I use reach to define how much a given lens magnifies your subject. I'm all ears for another way to define this. A TC adds reach at the cost of FoV. How would you put it?

[moxfyre]

Of course it narrows the FoV. So does using a 400 instead of a 200, or a 200 x2 via TC.

I define FoV as the width of the your field of view expressed as an angle. I use reach to define how much a given lens magnifies your subject. I'm all ears for another way to define this. A TC adds reach at the cost of FoV. How would you put it?

I would put it as "a TC increases magnification and narrows FoV". Reach=magnification, as you put it... but others seem to think of reach as something else so I get confused

[thePiRaTE!!]

Admittedly, I freely change reach, magnification and even (god forbid) zoom in there sometimes. I'm home schooled on this, what can I say.

You know reading what you and I said above, sensors aside its the same but worded differently -

...Nobody jumps up and down because 20mm now looks like 30mm. In fact, it still looks like 20mm (...ie magnification is a constant, ed), just with less image on the sides. The center isn't magnified, its just the sides are gone. Same for the long end...

My previous route is circuitous without the context of trying to unbrainwash APS marketing. Let me try this again with one set of terms - I'm saying that magnification being equal, the lower FoV of the APS-C gives you a cropped picture vs a FF.

...20mm on an APS-C sensor will not look like 20mm on a FF sensor if the APS-C sensor has the same number of pixels. It will capture as much detail as 30mm would on the FF

We are saying the same thing.

The rest of the science has been a big (albeit interesting) detour from my perspective, as this is all I was trying to say! I'll try to keep 'keeping it simple', simpler. (Was that irony?)

[aegisphan]

I basically quote myself in my previous post on my definition of "reach." Reach is basically the subject projection on the film/sensor.

FOV is dictated by the sensor size and the magnification/reach. FOV is basically the relative size of the subject projection and the whole frame. So if you take a same magnification on a 100mm and 10mm, you will still have the same FOV.

So the reach of a lens will not change regardless of the format use (35mm, 6x6, 645, etc.). What changes is the FOV, which, in this case, is controlled by the format size.

The subject's magnification is not the relative size of the subject projection and sensor size; it's the size of the subject projection. Since we are talking about the projected image on the sensor/film, we have to assume that the registration distance has to be the same.

At the same subject distance, the 300mm has more "reach" or projects a bigger image on the sensor than the 200mm. So if both lenses have the same resolution, the subject is now covered with more lines, and thus has more details.

I guess "reach" is more of a layman term. I used it because some of the materials I read used this term.

[moxfyre]

I basically quote myself in my previous post on my definition of "reach." Reach is basically the subject projection on the film/sensor.

But... what... does... that... MEAN???? "The subject projection of the film/sensor"? So reach means... the image formed by the lens??

Argghgh

FOV is dictated by the sensor size and the magnification/reach. FOV is basically the relative size of the subject projection and the whole frame. So if you take a same magnification on a 100mm and 10mm, you will still have the same FOV.

So, you're defining FoV as the linear dimensional extent of the scene in object space??? That's not in line with the usual definition at all: Field of view or angle of view is normally defined as the angular extent of the scene's projection onto the sensor. It can vary independently from magnification.

With the standard definition, FoV is uniquely determined by focal length, f, and sensor dimension, d: FoV = 2 arctan (d/2f).

And magnification is uniquely determined by focal length, f, and camera-to-subject distance, do: m = f/(f-do)

So the reach of a lens will not change regardless of the format use (35mm, 6x6, 645, etc.). What changes is the FOV, which, in this case, is controlled by the format size.

What? The actual, optical focal length stays constant when you put the lens on another format size. It is just about the ONLY thing that stays constant. So if "reach" stays constant, then what is reach... is reach the same as focal length????

EDIT: Hey, this is amusing. I googled "definition of reach in photography" and the first thing I got was another thread, on another forum... which degenerated into an argument over what "reach" means, if anything: Crop Body Madness - Page 6 - Canon Digital Photography Forums

[aegisphan]

But... what... does... that... MEAN???? "The subject projection of the film/sensor"? So reach means... the image formed by the lens??

Ya the image/projection that is formed by the lens on the sensor.

But... what... does... that... MEAN???? "The subject projection of the film/sensor"? So reach means... the image formed by the lens??

Argghgh


That's not in line with the usual definition at all.

Field of view or angle of view is defined as the angular extent of the scene's projection onto the sensor. It can vary independently from magnification.

Field of view is uniquely determined by focal length and sensor size.

Magnification is uniquely determined by focal length and camera-to-subject distance.

First off, I want to say I am too good with Physics, and Optics happens to be one of my weaker subject. I'm a Biologist. However, based on what I read and real-life observation, I form my own hypothesis. And that's what I'm presenting here . A hypothesis can be right or wrong given enough evidence. So that's what we're trying to see here.

Each focal length has its unique reach/magnification. Case in point, at the same camera-to-subject distance, a 200mm lens can magnify a subject 50m away at 1:10 size (not accurate number), while a 300mm would magnify the same subject at 1:8 size.

That's basically what you said: "magnification is uniquely determined by FL and CTS distance"

However, I don't agree with FOV is determined by FL and SS. Given the same SS, FOV can be the same for a 100mm lens and 10mm lens if the magnification of a subject is the same in both case. For example, taking a flower at 1:1 regardless of the focal will give you the same view.

So what I see is that, if you have the same magnification of a subject, the FOV will be the same regardless of the focal. That's why I said FOV is determined by magnification and SS.

What? The actual, optical focal length stays constant when you put the lens on another format size. It is just about the ONLY thing that stays constant. So if "reach" stays constant, then what is reach... is reach the same as focal length????

Sorry, I got it wrong right there, since I forgot about the registration distance.

At a given registration distance, each focal can cover a certain image circle. At the same registration distance, reach will be the same for a given focal regardless the size of the sensor. Of course with bigger sensor, you will have vignetting/light fall off.

If you can reason it otherwise, of course, I'm all ears.

[moxfyre]

Ya the image/projection that is formed by the lens on the sensor.

Okay, so... to you, "reach" simply means "the image projected onto the sensor"???

First off, I want to say I am too good with Physics, and Optics happens to be one of my weaker subject. I'm a Biologist. However, based on what I read and real-life observation, I form my own hypothesis. And that's what I'm presenting here . A hypothesis can be right or wrong given enough evidence. So that's what we're trying to see here.

I gotcha! The thing is, these are almost purely definitional issues... if everyone uses the same definition of the terms, the consequences are straightforward

Each focal length has its unique reach/magnification. Case in point, at the same camera-to-subject distance, a 200mm lens can magnify a subject 50m away at 1:10 size (not accurate number), while a 300mm would magnify the same subject at 1:8 size.

That's basically what you said: "magnification is uniquely determined by FL and CTS distance"

However, I don't agree with FOV is determined by FL and SS. Given the same SS, FOV can be the same for a 100mm lens and 10mm lens if the magnification of a subject is the same in both case. For example, taking a flower at 1:1 regardless of the focal will give you the same view.

You say "the same view" because you're defining "field of view" as something like "the width and height of the scene in inches/meters/whatever." Whereas, as I said above, the STANDARD definition is in terms of the *angle* subtended by the scene, as in "12 degrees wide by 8 degrees high." If you define FoV in terms of lienar dimension, your conclusion is correct... but that is absolutely not the standard definition, so it will confuse everyone

Also, taking a flower at 1:1 will NOT give you the same picture regardless of focal length!! If you do it with a 300mm macro lens, you'll have to put the camera about 5 feet away. If you do it with a 50mm macro lens, you'll have to put it about 1 foot away. The perspective will be completely different... the flower will look "flattened" with the telephoto lens, whereas it will look more 3D with the 50mm lens.

At a given registration distance, each focal can cover a certain image circle. At the same registration distance, reach will be the same for a given focal regardless the size of the sensor. Of course with bigger sensor, you will have vignetting/light fall off.

If you can reason it otherwise, of course, I'm all ears.

See, now that I know you define "reach" simply as "the image projected by the lens" ... I agree