A note on magnification, resolution and macro for digital

A forum to ask questions, post setups, and generally discuss anything having to do with photomacrography and photomicroscopy.

Moderators: MacroMike, nzmacro, Ken Ramos, twebster, S. Alden

Locked
_sergey_
Posts: 41
Joined: Fri Jan 28, 2005 3:34 pm
Location: Tver, Russia

A note on magnification, resolution and macro for digital

Post by _sergey_ »

Hello!

Sorry if you are tired of those endless discussions, but I believe that I realy see some important ideas here that are often overlooked. Anyway, the definitions of terms are important, because if we do not have definitions we will never understand each other. Misunderstanding of those concepts can also lead to financial looses, if one is selecting a "best" camera to buy, for example. So, I think all above excuses my very long post below :oops:.

First of all, I’ll show that we really have a problem. Suppose you have two digital cameras. First one have a 1:1 lens, sensor size of 2 inches and resolution of 5MP. Another doesn’t have such a good lens, only half life size. But it has another sensor – it is only 1 inch in size, but 5MP too. Now imagine you are shooting the same scene with both cameras at their closest possible distance. What will you get? You will get exactly the same image (well, we are not discussing amount of noise vs sensor size here)… What’s the point of having 1:1 lens vs 2:1 lens than? Of cause, in real life we will not have such an extreme example, but the problem is still here.

To see the reasons for this lets first return to the old film cameras and analyze why do everyone use image size at film plane to judge macro possibilities. Remember the idea Danny likes to tell us – 'care about images ...'. So, first of all I’ll say that what we should be really interesting in is a magnification achieved in our final product. What is a final image when we are working with film camera? Nowadays it is often a print. This is not a good thing for our analysis, since we can print our image any size we want. In the film world there is no exact limit of how much we can enlarge our print. Of cause, when we enlarge, we are loosing some quality, but the margin of 'acceptable' quality is highly subjective. For example if we will start seeing grain than we can just take a better quality film (or wait until technology will make it available). We can get better enlarger lens and so on. So, if we are going to make a comparison of different cameras than we should compare images enlarged the same number of times – because in this case we will have equal conditions for both cameras. So, what will define the magnification of the final image in this situation? The image size on the film plane! In some cases that image will in fact be your final image – for example if you are shooting slide film or use old large format camera.
Now we see, that 1:1 lens for medium format and 1:1 lens for 35 mm, and for any other film size really have the same 'macro power', they just have different field of view.

Let’s now consider digital camera. Here the situation is very different – for a digital image there is an optimal size to print or view. To get maximal quality you should print it so, that one pixel of digital image corresponds to one pixel of you output device. This rule fit well such output devices as monitor, but may be not that clear for devices like inkjet printer. The problem with inkjet printer is of another nature – the reason is that it is not capable of producing enough different colors at maximal resolution, so effective resolution for full-color image is much less than one shown in specs. Anyway, I’ve read some books about preprocessing of digital images for publishing, and all of them say the same thing – 'resolution of your image should match the resolution of the output device'. So, lets consider such output device as monitor, because it is simple and known for all of us. Let’s calculate the magnification of the image we are seeing on our monitor vs size of the ant we have found in a back yard. Let it was an ant of length L inches. Size of image at the sensor plane of digital camera will be L*K inches, where K is a magnification power of a lens. A sensor with resolution of S dots per inch will create digital image of the ant, consisting of L*K*S pixels. When we are seeing this image on screen at resolution which equals to resolution of our screen D dots per inch, we will see an image of size L*K*S/D inches. So, what magnification do we get? It is (L*K*S/D)/L=K*S/D. Since we 1) can have different screens with different resolution 2) should take other output devices in consideration, which will have other D -> we should consider D to be some constant value and ignore it (like it was with print magnification in the film case). So, the magnification of our digital image is characterized by the K*S (lens magnification*sensor resolution). In many cases in practice we do not have access to the values of K and S. But we can measure their product, because it is minimal length of an object at subject space, which corresponds to one pixel, expressed in mm/pixel or inch/pixel which one you like. So, if you ask to include resolution to the magnification equations, here it is.

Theoretical background.

When I’m saying that lens magnification alone cannot be used to judge macro possibilities of digital camera, I’m not going to break laws of physics. It just reflects the fact that images of film and digital camera are of a very different nature.

Analog (film) image should be considered continuous, in the sense that we could (theoretically) extract that much information from it, how we need. Another important property of the film image is that it really has 'the size of image at film plane'. Macro possibilities of film camera can be measured by comparing lens magnification/or image size at film plane.

Digital image is discrete – it only has the amount of information recorded in its pixels, and not more. And this should be taken into consideration when you are comparing the digital cameras (for macro work or for any other reason). Another important property of digital image is that it does not have the physical size in the sense the film image has it. Note, that in the example above size of digital image was giben in pixels, not inches or meters! You can imagine digital image as an infinitely resizable mesh – your can stretch or expand the mesh to the size you want and than fill each cell (pixel) with some color. Pixels of digital image do not have physical size when the image is stored inside your camera or computer. Pixels do have size only when your are converting analog image to digital (inside camera or scanner) or digital to analog (printer, monitor), and this process is characterized by the resolution. You can freely assign any size to the pixel; this will not change the image itself. That’s why it does not take much time for graphical editor to change image resolution – it just changes single number in the file. In fact, images inside computer are resized many times even in cases when we do not think about it – when we are viewing an image on screen our favorite image viewer just ignores information about image size stored in file and uses the value appropriative for you screen. So it just does not have any sense to use 'image size at film plane' for digital image, since digital image doesn’t have any size at all. To analyze process of capturing digital image (including analyzing macro possibilities) we should use resolution.

Practical conclusions.

1. If we are going to compare film cameras we should use lens magnification/image size at film plane.
2. If we are going to compare 'Point and Shoot' type digital cameras the easiest way is to use resolution.
3. If we are going to compare different lenses for the same digital SLR (or for SLRs with the same sensor size and resolution) we should use lens magnification.
4. If we are going to compare digital SLR and point and shot camera, or DSLRs with different sensors we should calculate resolution achieved by SLR using lens magnification and sensor resolution.
5. If we are going to compare digital and film cameras … we should give up. There is no way to compare them, because resulting images (film and digital) are of incomparable nature. In English they say 'it’s like comparing oranges and apples'. It is just our case – you can’t divide an orange which have 14 'predefined' sections for 15 people, but in the case of apple you will just cut slices a bit smaller. And our digital oranges are of imaginary nature and do not have size!
Well, you can do some comparison, but it will not be fair – it should include your subjective impressions of what you consider 'acceptable' quality or 'resolution' of film image. For example I rarely print images larger than 15x20 cm, so I can compare magnification achieved at this print size, but if you print 30x40 than you will have different result.

Definition of term 'macro'.

Now we have another problem. In film times we define macro to be started at life size. Now you should see that we can’t use this definition directly (because we do not have size) and, to make things worse, we do not have a way to translate this definition to digital terms (explained in 5). So, people, it is a good time to find completely new definition for macro. It should be expressed in resolution terms. Where to start it? I do not know… In film world we have a 'appealing' margin at 1:1 (note that nobody really knows the upper limit!). So may be it should be of the same 'appealing' nature too, something like 1 pixel/mm, 100 pixels/inch, etc…

Hope at least someone will read it to the very end and understand it... Thank you.

WBR,
Sergey.

User avatar
wilash
Posts: 104
Joined: Wed Apr 13, 2005 4:43 pm
Location: Japan

Post by wilash »

Sergey, I think you are thinking too hard. (Either that I missed your point.) There are so many ways of comparing imaging systems that to define the method before deciding on the context seems limiting and may not be practical. Any comparison needs to be qualified and their is nothing wrong with your guidelines, but they are not absolute.

Knowing the exact degree of magnification of an imaging is only important if the image will be used in measuring scale. Even with the best systems, there is a degree of error - do you know the actual focal length of your optics? But I see many people here who are simply interested in making aesthetic images. That is also fine. So in that case, the exact reproduction size is irrelavant. Systems could be compared by aesthetic responses - color, contrast, etc. Or simply if they can get close enough.

Digital cameras do present a problem. One is how reproduction is measured. Some manufacturers use the term "life size" to indicate the area covered by their camera as if it was made by a 35mm camera. As you can see, that is a little confusing. A kind of 35mm marco equivalent. Currently there are no standards in digital imaging to describe macro capabilities - I know there are ways of doing it, but the manufacturers are not bound by them. (So be careful of that 1:1 and 1:2 claim. Check if they state the area covered at the minimum focusing distance.)

BTW, enlarging images should not introduce a loss of quality when view at the correct viewing distance. There is no limit to the size of enlargement that can be made from a given image (except a technical one). An image is no more grainy at 8x10 than 20x24.
Will

User avatar
twebster
Site Admin
Posts: 1518
Joined: Mon Apr 19, 2004 10:55 pm
Location: Phoenix "Valley of the Sun", Arizona, USA

Post by twebster »

OK, folks :!:

1:1 reproduction is life size. a 1/2" fly will measure 1/2" whether it is on a 35mm negative, a 2/3" ccd, or on a 4" x 5" piece of film. Period. Life size is life size. If that 1/2" fly measures anything but 1/2" on any piece of film or ccd/cmos sensor, then the reproduction ratio is not life size. Period. It doesn't matter the format of the recording media. Image magnification measured at the film/ccd/cmos plane determines the magnification of the image. If that 1/2" fly measures 1/4" on the film or ccd/cmos sensor, regardless of the size of film or ccd/cmos sensor, then the magnification is 1:2 or 1/2 life size. Period. If that fly measures 1" on the film or ccd/cmos sensor then the magnification is 2:1 or twice life size. Period.

This ain't rocket science people...Information like this gets distorted because the basic laws of light physics are ignored. Information like this gets distorted because people believe everything that is in print, whether it is correct or not.

I don't mean to offend anyone here by all means but misinformation is misinformation and does nothing more than to add to the already general confusion regarding photography and optics.

Btw, pixel density has nothing more to do with image magnification than fine-grained film vs. coarse grained film. Circles of confusion are circles of confusion (those make up images BTW) and it don't matter on what you record those circles of confusion. Period.

Sorry if I offend anyone with this response....

Sorry, this statement is false, "as if it was made by a 35mm camera". This is not true. Manufacturers measure life size as being life size. Period.
Tom Webster
Administrator

Phoenix "The Valley of the Sun", Arizona, USA

Think about this...maybe Murphy is an optimist!!!

User avatar
wilash
Posts: 104
Joined: Wed Apr 13, 2005 4:43 pm
Location: Japan

Post by wilash »

I work for a manufacturer which makes, among other things, digital cameras. The statement I made was from experience. What you say is true, Tom, but that does not mean manufacturers can't or don't bend the rules so to speak. So it is best to check the product specifications carefully when it comes to digital technology.
Will

_sergey_
Posts: 41
Joined: Fri Jan 28, 2005 3:34 pm
Location: Tver, Russia

Post by _sergey_ »

Hello!

The main points of my post were:
1. Lens magnification alone (!) is not sufficient to judge macro possibilities of the digital camera
and
2. If one wishes to select a digital camera which allows to get "closer" than other it should pay attention to resolution, like it was with magnification with film camera.
3. I'm interested in the final product, which is an image. I'm not interested in lens design, CCD design or any other things by itself, I'm only interested how do they influence final digital image.
1:1 reproduction is life size. a 1/2" fly will measure 1/2" .... a 2/3" ccd
Tom, what I was trying to say is that you will not get any benefit from this fact. If you are shoting film I see that having 1/2" fly on film is much better than 1/3". But for digital, I do not see why it will help anyone to see smaller details of fly, or any other difference for the image you get.

You see, while light passes different groups and lenses inside you lens several images are created. Do anybody know what is a size of image created by the front group, for example? Is it lifesize, greater or smaller? Nobody but optical engineers care about this, because for practice only final image is what really matters for photographers. For digital image you have both lens and CCD before you get an image, and both of them have equal influence on resulting digital image.

That man who wrote an article at
http://www.cs.mtu.edu/~shene/DigiCam/Us ... ation.html
really do strange thing: he takes several cameras, for example Nikon 5700 and Nikon 5000 and he concludes that the first one is better suted for macro photography because it lens has 0.36 magnification instead of 0.31. And this is a very wrong conclusion. I do not realy know which one is better, but if the latter one has more CCD points it can be better. And this guy look like someone who tells that one lens is better than other because greater magnification is achived after the first group inside the lens.

To make things worse manufacurers (who whant to sell their cameras first of all) do show many different numbers which are supposed to show us that their cameras are better (let's say for macro). They can use "magnification at 35 mm eqivalent", "magnification", "minimal filed of view" and a lot of other things. I am a sort of person who really read a lot of specs before investing money to be sure that the product I'll buy will fit my requirements... So, I read all those numbers, compare them.... And than I understand that most of them do not mean anything for me! And I was able to cleary see it comparing images itself (may be all this will be much more clear if I add Illustrations? Don't be afraid, I wont do it here).
You are right, Wilash, we should be very careful with those specifications, and we should think what do they really mean. Resolution is what really matters if you are interested in seeing smallest details.
Information like this gets distorted because the basic laws of light physics are ignored
I do not ignore the laws of light physics. Yes, magnification of lens is important, but for digital it is not everything. If you see that I ignore something, please show what it was.
Btw, pixel density has nothing more to do with image magnification than fine-grained film vs. coarse grained film.
I've shown by the equations, that it really has something to do with magnification of digital image.
Digital image do not have other size than pixel count, and pixels do not have width and height. After you take the image with CCD it does not matter how large were pixels of that CCD. Were they big or small, they are all the same inside computer. But if you take image on coarse grained film, it will be image with coarse grain, was it at 35 mm camera or at medium format.
And pixels are aligned in regular fashion. All this is very different from film grain, and it has much influence on what you can do with the image, try to understand it... You see laws of physics, but not laws of computer scienece.

Yes, some lens may be sharper than other and have smaller circles of confusion, but I was not intended to discuss lens quality. This way one can tell that 1:1 lens for 35 mm is better that 1:1 lens for medium format, because it is easier to desing sharper lenses for smaller film size (I've seen people saying things like this).

I hope circles of confusion are much smaller than pixel size of you CCD, because otherwithe you will not be able to realise it full potential (resulting image will not hold all the information it can). But if this happen, than we just have a poorly designed camera, and nothing more. Note, that 1:1 magnification of lens do not mean it has good quality at this magnification.
BTW, enlarging images should not introduce a loss of quality when view at the correct viewing distance.
If anyone is going to see smallest details possible (that's the whole deal in macro) than he will try to enlarge image as much as possible and will view it from close distance. This will amplify grain (from original film, not print grain), circles of confusion of original lens and so on. So you large image won't look that sharp as a small one. This is what I was trying to tell.

P.S. When I was selecting a camera I wanted one that will allow me to take better macro photos (well, among many other things :) ), I wanted to be able to take photos of smaller objects than with equipment I have before. What is a better macro photo? That one where you can see smaller details. How can one measure smallest detail you can capture with digital camera? With resolution... If this is not clear too, I'll give up :). And hope someone else will understand. I won't post more on this subject here. If anyone is interested in the discussion feel free to contact me at sergey@tversu.ru.

P.P.S. And old definition of macro will die... Just to many cameras which are able to take photos which look like macro, and none of them fits this definition. It will not work. Period. And this is a law of pshychology, I suppose :). It even does not matter who is right or wrong.

WBR,
Sergey.

User avatar
wilash
Posts: 104
Joined: Wed Apr 13, 2005 4:43 pm
Location: Japan

Post by wilash »

If anyone is going to see smallest details possible (that's the whole deal in macro) than he will try to enlarge image as much as possible and will view it from close distance. This will amplify grain (from original film, not print grain), circles of confusion of original lens and so on. So you large image won't look that sharp as a small one. This is what I was trying to tell.
But the correct viewing distance of the print is proportional to its dimensions. Since the angular resolution of the eye does not change and the viewing distance is propotional to image size, there is no difference in the percieved quality. If you look at a print closer then the correct distance, quality changes based on the change in proportions of the correct distance not the print size. A 8x10 print at half the correct viewing distance, looks as grainy as a 20x24 print at half the correct viewing distance.

There is no limit to the size of the print from a digital or film image. There is no loss of quality either. The precieved quality of a print is a product of where you stand.

I know this sounds strange. Many people will talk about reproduction limits, but it is really upsidedown reasoning.
Will

User avatar
wilash
Posts: 104
Joined: Wed Apr 13, 2005 4:43 pm
Location: Japan

Post by wilash »

P.P.S. And old definition of macro will die... Just to many cameras which are able to take photos which look like macro, and none of them fits this definition. It will not work. Period. And this is a law of pshychology, I suppose . It even does not matter who is right or wrong.
Well, this is true. Partly because the definition are arbitrary (i.e. based on a pschological response) from the beginning. This is why context is important for any definition. Just try getting scientists, painters, and photographers together to talk about color. :D
Will

User avatar
wilash
Posts: 104
Joined: Wed Apr 13, 2005 4:43 pm
Location: Japan

Post by wilash »

Btw, pixel density has nothing more to do with image magnification than fine-grained film vs. coarse grained film. Circles of confusion are circles of confusion (those make up images BTW) and it don't matter on what you record those circles of confusion. Period.
I am going to have to agree with Sergey on this. Since angular resolution and resolving power is directly related, it is quite possible to to equate the two. Whether it is necessary or practical is another matter. It really cannot be compared to grain as pixels have a definite spatial position were grain does not. Pixel resolution is really not the same as granularity.

While the pixel resolution may indicate magnification, it does not describe the actually image resolution as it does not take into effect the lens resolution. A problem of the "optical" resolution of scanners which is only describing the pixel resolution and the spatial resolution of the stepper motor. The resolution of the scanned image is always something less.
Will

User avatar
nzmacro
Site Admin
Posts: 1604
Joined: Tue Apr 20, 2004 1:25 am
Location: New Zealand
Contact:

Post by nzmacro »

Oh boy, been down this track many times :wink:

Its all CLOSE UP :wink: . I mean heck, Nikon called their macro lenses Micro and Tom and I have seen where its been argued that it was because Nikon had larger ratios than Canon :D .............ya what :shock:

Macro, 1:1 or larger on film size or sensor. Micro = image taken through a microscope.

Thats how I've always seen it and regardless of ratio above 1:1

As for printing, naaa, I'll just use a large computer monitor to get a larger ratio :wink: :D

Danny.

gunn
Posts: 393
Joined: Tue Jan 04, 2005 2:59 am
Location: Adelaide/KL

Post by gunn »

Hello Sergey

I do admire the passionate intensity with which you seek to expound this volatile subject-matter. At this juncture, I feel constrained to make two points:
"That man who wrote an article at
http://www.cs.mtu.edu/~shene/DigiCam/Us ... ation.html
really do strange thing: he takes several cameras, for example Nikon 5700 and Nikon 5000 and he concludes that the first one is better suted for macro photography because it lens has 0.36 magnification instead of 0.31."
(1) With respect to the above quote, I fear I am unable to arrive at a similar conclusion as you had from my reading of the said article. I have read and re-read the article, and have no difficulty in surmising that the author was merely comparing the images from the different digicams and calculating their different magnification ratios. It did not, and still does not, seem to me, at least in spirit, that he was asserting that a digicam with a higher magnification ratio has, per se, better macro capabilities than one that has a lower ratio.

(2) I think Tom here is of the opinion that pixel density on a sensor amounts to no more than grain density on film. I tend to agree. This naturally, therefore, begs the following question: Are you (Sergey) saying that iso100 film is macrographically superior to iso400 film? Or are you saying that there is a distinction between film-grain-coarseness and film-grain-density.

I shall not be responding further in this thread. I only felt constrained to respond because I had pointed out the article's URL to both you and Tom. I hope we find the Grail one day.

Best wishes and regards
Benedict Gunn

Locked