Joined: 01 Aug 2006
Location: Richland, Washington State, USA
|Posted: Tue Jan 01, 2008 9:28 pm Post subject:
|I see that Will has replied to Charlie while I've been typing this. I'm glad to hear that we all now seem to be on the same page. But I'm going to hit Submit anyway, just to capture these thoughts.
Those are excellent links, Charlie. They definitely show good clear ways to denote magnification without having to spoil the aesthetics of an image by putting a scale bar in it.
I think it's also worth pointing out that these are SEM images. They were created directly in electronic form by monitoring temporal fluctuations in electrical current while a very small and tightly focused beam of electrons was scanned across the subject.
For these images, there never did exist any spatial sensor at which it would make sense to specify a magnification. Microns per pixel and pixels per micron are the natural units of the instrument. Magnification is something that gets figured out at the end, as one of two ways to communicate with the viewer after the image has been converted to physical form.
I use the same model for digital photography. In that model, sensor size and magnification on the sensor are not relevant, except as intermediate numbers on the path to what I really care about: "How big is this subject, anyway?".
As it happens, I have in hand some SEM images of the yucca moth that I documented here. Those images were provided to me as TIFF files, written directly by the instrument that created them, and sent to me by the seasoned professional who operated it. The images carry both scale bars and nominal magnifications such as "50x". It is interesting to analyze and compare those notations.
The scale bars are straightforward. They visually indicate measurements such as 500 µm (0.5 mm), and they agree completely with the scale bars that I constructed by just image manipulation -- no calculations -- using the technique described here.
The nominal magnifications are less straightforward. No, that's not right. Actually they're actively misleading.
The file in front of me right now, L4437-1.TIF, says "50x". The image size, according to information embedded in the file as reported by Photoshop and IrfanView, is 1228 x 960 pixels, 24.3 x 19 inches. So the calculation is straightforward: the subject width must be (24.3 inches * 25.4 mm/inch) / 50X = 12.34 mm wide. But this number is wrong. According to the scale bar, the frame width is actually 2.27 mm wide.
In other words, the stated magnification, combined with the stated format size, gives an answer that is incorrect by a factor of over 5X!
What's going on? I have no idea. If I ask the question, "How big would the print have to be, in order for 50X to be the correct magnification?", then I get an answer of "About 4.5 x 3.5 inches." So I suppose I can speculate that the manufacturer and/or operator of this equipment have standardized on magnification with respect to a 4.5 x 3.5 inch image. But they don't say that, there's no way for me to know that, and the information embedded in the image file says something completely different.
Of course it's easy to dismiss cases like this by saying "Obviously these people didn't understand what they were doing. They need to be educated better." I won't disagree with that interpretation. Better education is always good.
But part of that education, it seems to me, should be understanding the tradeoffs of various methods. Specifying "magnification" can be an effective way to communicate intuitively, but it is terribly vulnerable to becoming misleading or incorrect due to lost, corrupted, or misapplied information. When in doubt, trust the scale bar.
As I've stated above, my preference is to have both, and to spell out explicitly what my "magnifications" mean.