Getting started

[grepper]

Yes, Herman.  That is exactly what I'm trying to say.  I don't see how you can infer anything as far as magnification from measuring or comparing images on a particular display device without being able to compare it to a 0.0X magnification image of the same object, or at least a 0.0X magnification of some subject of known dimensions.  Obviously an image displayed on a 50" TV is going to be much larger than the same image display on a phone.

Jan- Please ignore my previous post:  I reread that sentence in your post and now understand you were just showing an example of something 85x larger than something else and not trying to infer anything as to the magnification of the scope.  Doh!  Sloppy reading on my part.  My apologies.

Obviously, I could take two images with the same device, at the same magnification, and knowing the dimensions of an object in one, determine the size of some object in the other. 

If I could take an image with my scope at 0.0X magnification I'd be good to go.   Barring that, all I have is a bunch of images at an unknown magnification.  Hopefully I'm not missing the forest cuz of all the trees.

Herman – What is your guesstimate as to how those images compare to what you see @ 40X?

I do know the image dimensions, and this display is 1280 pixels horizontally, and the physical display size is ~304 mm horizontally.  So, about 4.2 pixels/mm.  From that I could approximate the pixel size of this display.  I know the approximate field width of the scope on max magnification from the image of the ruler.

Maybe there is something in that from which I could somehow figure the magnification by comparing a displayed image to physical reality.  But that that would not be fair!  It would require thinking.  Nobody ever told me that sharpening would require actual thinking!

[grepper]

Here is a better image for comparison. It is how my scope at full magnification sees the date on a 1968 penny.

Jan- can you post a similar 50X and 100X image?  That would be a cool comparison...  Crap!  I wasn't thinking...  You live across the pond from here.

A riddle:  What small microscopable object do we have in common?

[Jan]

It can't be as simple as just holding a ruler up to the screen and measuring the distance of one displayed mm in an image to one mm on a physical ruler.

It is if, as Jan said, it's the magnification on your monitor. That tells you nothing about the magnification of your scope.

To see this take a photograph of some object with a camera and then carefully scale it so that when you print it the image is the same size as the object. Then post that image on a web site. When others look at that image they will see it with different magnifications. Some will see it reduced in size because they're viewing the web page on their phone. Others will see it enlarged because they're viewing it on a projector screen.

You are correct, Herman. 

Where the magnification (scale) of an image is important we usually superimpose a scale bar on the picture. The scale bar has a stated length and is resized together with the picture. From a picture with scale bar we can easily calculate the actual magnification regardless image reduction or enlargement. 

Jan

[Jan]

Obviously, I could take two images with the same device, at the same magnification, and knowing the dimensions of an object in one, determine the size of some object in the other. 

I think you are by leaps and bounds approaching full understanding of the problem. I am very happy about it. 

Jan

P.S.: When you are speaking about 0.0X magnification you have probably in mind 1.0X magnification, because 0.0X means zero apparent size of an object.

[Jan]

Here is a better image for comparison. It is how my scope at full magnification sees the date on a 1968 penny.

Jan- can you post a similar 50X and 100X image?  That would be a cool comparison...  Crap!  I wasn't thinking...  You live across the pond from here.

A riddle:  What small microscopable object do we have in common?

Mark, attached you can see a hair of my wife, mm marks of a steel ruler and a date on a 1 Euro cent coin, dated 2009 (German edition: Ein Euro Cent).All images were taken with optical magnification 50X and resampled from 3264x2448 to 480x360 pixels.

The calculated hair diameter is circa 70 μm and the calculated height of the figures on the 1 Euro cent is almost 1 mm.

Jan

[Ken S]

Jan,

My older eyes have trouble seeing things at 0.0X magnification. Could we try to enlarge things just a bit.....

Ken

[Jan]

You overestimate me, Ken. In this country we have a saying: "Where is nothing, not even devil can take." 

Jan

[grepper]

What???

0 * 5 != 5 
5 * 1 != 6 

[Jan]

Mark, I have reread your post from yesterday and I can tell you that many of your words and sentences are clear and sharp as a two-edged sword piercing the details of magnification definition. 

The point is, that magnification is defined as a ratio between the apparent size of an object (size in an image) and its true size. All multiplicative factors you have mentioned (e.g. PPI), are in our two images approach, both in the nominator and denominator and therefore do not influence resultant magnification. They cancel out each other. 

A famous quote attributed to Leonardo da Vinci says: "Simplicity is the Ultimate Sophistication". 

Jan

[Herman Trivilino]

Herman – What is your guesstimate as to how those images compare to what you see @ 40X?

Here's the thing, since your magnification is entirely digital, it doesn't matter. You can always look at them on your computer and zoom in to your heart's content. Without optical zoom, though, you won't get any better resolution, meaning you won't see more detail.

I can tell, though, that you have a very narrow field of view. Just 2 mm. I can see about a quarter of a penny.

[grepper]

Herman, Yup.  I understand optical magnification.  Absolutely no problem there.  I also understand that I can take a displayed image captured with no magnification  and compare it to an image captured at X optical magnification displayed on the same physical device, and determine the amount of magnification simply by measuring the display.  That's not what I was getting at.

I also fully understand optical magnification vs digital zoom.  Digital zoom on the capture device is, IMHO, less than worthless.  You can always zoom later on the computer.  In fact, depending on the complexity of the scaling algorithm, it can actually be misleading due to interpolation.  Many sophisticated scaling algorithms will interpolate data so as to smooth edges, fill in holes, perform color equalization and reduce JPG artifacting so that the end product is more pleasing to the eye.

This can be desirable in photography when cropping and then enlarging an image for printing.  Some of the sophisticated resizing applications such Perfect Resize specialize in it, and the scaling algorithms in Photoshop and the like do a most excellent job.  It's rather amazing how much scaling you can do and still produce a reasonably good looking image.  And that's not even considering the almost magical modern inkjet printheads, their control software and associated hardware, but I'm babbling. 

But, you sure wouldn't want any of that "helpful" interpolation if you were looking for small anomalies in cell structure for example.

That is not the issue I was getting at.   What I'm still trying to understand is scaling, either up or down, inherent in displaying an image on a particular physical display device with a physical pixel density and size.  For instance, you cannot display a line with a thickness less than the physical width/height of a pixel on a display if you define a pixel to be the combination of its RGB elements.  If you only illuminate the R for example, is the displayed pixel 1/3 the size?  Not sure about that.   Nonetheless, even if you could display a finer line with, say, color #0,0,FF, the closest object that could be displayed next to it would be one RGB pixel element away in which case it would appear to be a solid anyway.  I'm rambling again.

Actually I happen to know that with computers at least all of this data is defined and accessible if you are willing to write some code or are using an application that takes advantage of it.  A displayed object size can be device independent as long as it is not so large to exceed the boundaries of the display or so small that it cannot be represented with a single physical pixel.

The only reason I started thinking about this was when Jan said:

Grepper, thanks for posting the image. If it is really 200X magnification than the edge is perfect. 
Jan

That got me wondering if, when displaying an image of unknown original optical magnification, it was possible to figure out what the original optical magnification of the capture hardware was, not including sizing due to the interpolation of digital display devices.

My microscope has calibration software, and according to it, depending on exactly how close the lens is to the subject, the magnification is between 180X and 200X.  The scope is rated at max 200X optical, focal length 5.8mm, FOV: 13.0°.  Its specs correlate with the reported values from the calibration software.  So guess what... I'm going to believe it.   The images I post are ~200X magnification.

[Jan]

Grepper, you have confirmed that you are the foremost microscopist of our forum, congrats. 

Having now all knowledge and understanding almost all mysteries, can you tell as what is the diameter of the splited hair shown in your reply #40? Was my row guesstimate 60 μm OK?

Jan

[Herman Trivilino]

What I'm still trying to understand is scaling, either up or down, inherent in displaying an image on a particular physical display device with a physical pixel density and size. 

The pixel size on the CCD is the limit of what you can resolve. You can magnify it all you want, but beyond a certain limit it won't help you identify anything in the image.

For instance, you cannot display a line with a thickness less than the physical width/height of a pixel on a display

I think you can. If, for example, a line is narrower than the pixel width it may in some cases show up on the image as a line whose width equals the width of one or two pixels.

[Herman Trivilino]

In Reply#54 I meant to say that I can see about a quarter of a penny on my scope. I think one of the designs of a dissecting microscope is a wide field of view.

[grepper]

Herman-  I think your dissecting scope is perfect for looking at blades.  It would be nice to have more magnification, but they have, like you say, a wide FOV and are set up for specimen lighting from above.  They also have room for tall subjects. 

The problem with biological scopes is that the objective lens must be very close to the subject and they don't have a lot of room for bulky subjects.  They are generally set up for lighting from underneath for viewing slides, and are difficult to light stuff from the top or side.  But they have higher magnification which is cool.

The problem with most USB scopes is that they are cheap, very fussy to focus, don't have a large FOV and the magnification changes depending on distance from the subject so you never know exactly what magnification the thing is set to.  You just turn it all the way to "+" and assume it is acting to spec.  Unless you spend a lot, they all have total junk for stands that makes focusing even more problematic.  The good thing is that you can light from the top and sides, and since they are just a tube with a wire sticking out of them you can place them on any size object.  Also, they are inexpensive and work good enough to be useful.

I think you can. If, for example, a line is narrower than the pixel width it may in some cases show up on the image as a line whose width equals the width of one or two pixels.

Exactly Herman!  That's what's got me scratching my head.  Even if the capture device was dense enough to resolve the line, the 96 dpi monitor resolution might not be able to represent it.  In which case the the magic software interpreting the camera data for display would display a thicker line.  Now multiply that by a 1920 horizontal pixel monitor display and the resulting error could be significant if you were trying to ascertain anything by measuring directly on the display.

Having now all knowledge and understanding almost all mysteries, can you tell as what is the diameter of the splited hair shown in your reply #40? Was my row guesstimate 60 μm OK?
Jan

Jan- "Having now all knowledge and understanding"...  Please stop.  Surely you jest.

Sorry, I can't verify the hair size, but I will calibrate and do another one. 

Why can't I verify the hair dimension?  Because since images must be scaled to be attached to a forum post due to file size limits, or even as a direct link to Photobucket so they won't appear giant when displayed in the forum, I used a process that was as fast as possible to get the job done, without even having to load the image into some program to resize it.  What I did was:

1.  Display the captured image in an image viewing app with the shrink to fit option ON, maintain aspect ratio ON, so that the whole image would fit in the program window.

2. Grab the corner of the window and shrink it to some small size say, ~400 pixels wide by whatever, so it would be smaller.

3. Take a screen capture of the now smaller window.

4.  Save the screen capture as a JPG file and either attach it to a post or upload it to Photobucket.

That was the quickest way I could figure out how to do it.  It only takes a minute.  While fast, it does mean that by the time the image is posted it has not only suffered scaling, but also JPG conversion with its associated "lossy" compression scheme with artifacting and image quality loss.  But... it's a very expeditious way to reduce a 14MB+ image to 100KB or so.

Anyway, I looked around for the original image and can't find it.  All I have is the little scaled ones that I uploaded.  I can't find the hair laying around on my desk either. The Web sez: "the diameter of human hair to range from 17 to 181 µm".  So you are right in the ball park.

As far as the whole magnification & displayed image thing is concerned, I don't even know if I'm asking the right questions or even what it really is I want to know!  All I know is that, even though I don't know exactly what, something bothers me, and I know I don't understand what it means to capture on a CCD of some dimension and pixel density and then display it on a LCD of some other dimension and density with layers of hardware and software in between.  While interesting, does it really matter?   No.

But, I think I found the answer! Or at least something poking around whatever it is I wonder about.  The only problem is that while stuff like the Nyquist-Shannon sampling theorem is apparently very explained, it is all so far over my head I'll never understand it.  Oh, well.

http://biology.stackexchange.com/questions/36528/calculation-of-final-magnification-when-using-a-ccd-camera

I'm just happy to know that the images I post are ~200X .