Gnurple dot Net

Since starting the Every Day Objects game (see the prior post), I’ve been doing a lot of macrophotography. My favourite lens set-up is an OM Zuiko 28mm f/3.5 lens, mounted on a 25mm extension tube (EX-25) using the OM-to-Fourthirds adapter ring. The one that Olympus used to give for free to the first three people who asked very nicely on the second Tuesday after a blue moon (or whatever their algorithm was).

I became curious about my lens setup. Is this really macrophotography? According to Wikipedia, “… the classical definition is that the image projected on the ‘film plane’ (i.e film or a digital sensor) is close to the same size as the subject.” I have read on other sites that true macrophotography achieves a 1-to-1 ratio or better. I find the latter definition more appealing to my inner geek than anything sporting an “… is close to …” so that’s what I’m going with. I’ll leave it to the reader to figure out whether “more macro” means a higher magnification - what a microscope does - or not. Heh.

Macro With the 28mm Lens

The OM Zuiko lens’s focus ring works by changing the position of the front element. How does this affect us in macrophotography? With the extension tube, this hardly changes where the focal plane is at all, but it has a big impact on the field of view.

[ASIDE: For macrophotography, we usually focus by moving the object or the camera closer or farther from each other, not by turning the focus ring. In the 3-5x range (objects being 3-5x larger on the image plane than in real life), you would mount your camera on a platform which you move by turning a long bar with a very shallow thread to achieve focus.]

This lens has macro range which we know so far to be between “close-up” and “even more close-up”. I want to replace those wishy-washy phrases with real numbers. Apologies to any readers who were hoping for imaginary ones, i haven’t got any (badum-bum-pish, I’m here all week, try the fish).

Determining Macro-Factor

In order to find our macro-factor, we need to know the size of our image plane and the size of our object. For the latter we’ll use a scale (what normal people call a ruler, but which a professor at university very effectively drove into my brain is called a scale) marked in millimeters.

For the former, we must harness the infinite power of teh intarwebs. The Fourthirds standard says the sensor is 18 x 13.5mm with an imaging area of 17.3 x 13.0 mm. Done.

Image A: Focus Ring at Infinity

Here’s my scale taken with the lens set to its widest field of view (infinity focus) and then focused by placing the camera lens parallel to the scale, and moving the camera back and forth to achieve focus. I find the most accurate way to focus is to use Live View with the view set at 10x magnification - I knew Live View would come in handy someday.

Looking at the picture, we can see it is about 20mm wide, but I’d like to try and make that a bit more precise. To do so, I overlay a 10 pixel by 10 pixel grid on the image:

10 Pixel Grid

Here is a 100% zoom of the rulerscale (dammit!), with a 10 x 10 pixel grid overlaid. NOTE that this is taken from the left-most edge of the closer image (see below), not the wider image above.

I then determined how many pixels per millimeter at the left edge, center, and right edge, to the nearest 10 pixels. In Image A it was 180 pixels per millimeter all across.

Finally, using the left edge of each nearest millimeter mark, I counted the number of pixels to each edge of the frame, to the nearest 10 pixels. On the left edge, it’s 30 pixels, and on the right edge it’s 30 pixels as well.

1mm / 180px * 60px = 0.333mm + 20mm = 20.3mm^[1]

Image B: Focus Ring at Closest

Using the same techniques on this image, I get:

* 210 pixels per millimeter
* 30 pixels on the left edge
* 70 pixels on the right

1mm / 210px * 100px = 0.476mm + 17mm = 17.5mm

Conclusions

For the widest field of view:
17.3mm image area / (1mm / 180px * 60px + 20mm) = 0.85x macro^[2]

For the narrowest field of view:
17.3mm image area / (1mm / 210px * 100px + 17mm) = 0.99x macro

*so close*… sigh.

Confusions, a.k.a. Why Your Results May Be Different

Looking at the dimensions of the pictures I posted causes some confusion: the specs say that the camera produces pictures with a resoultion of 3648 x 2736 px, but my full-size images are 3720 x 2800 px.

Whiskey Tango Foxtrot, say you?

Being an all-linux guy, I use dcraw to process my raw photos. Why is dcraw producing images larger than the specified resolution? Dave Coffin, dcraw author, offers this explanation in his FAQ:

Why are dcraw output images larger than camera JPEGs?

    Any algorithm that combines each pixel with its neighbors is going
    to have problems near the edges. C code is cheap, so dcraw applies
    a different algorithm to edge pixels. Hardware logic is expensive,
    so cameras crop off the edge pixels after processing.

To understand why edge pixels are different, we need to know that the sensor is actually made up of colour-insensitive photosites covered by colour filters. Usually this is the Bayer pattern - though I don’t know what the E-510 in particular uses.

The E-510 specs, for some reason, list the same resolution for RAW files as for JPEGs. I don’t know what resolution Olympus’ software or any other Olympus Raw Format (ORF) to JPEG converter produces.

If we had done this test using JPEGs instead of RAW files, we’d've gotten a range of 0.87x - 1.01x because we’d be assuming the 17.3mm wide image area is 3648 px across, un-cropped. Now we’re reasonably sure it’s at least 3720, but we can’t really be sure of anything - did Olympus publish the 17.3mm “image area” width as the width of the area where pixels are, or where the 3,648 pixels which get used for JPEGs are? Are there other pixels beyond the 3720 we know about?

Am I wrong? Is it possible I do actually have a 1.0x macro factor? My measurements would have to be off by more than 30 pixels, or the width from Olympus by more than 0.05mm. Despite my poor technique and non-horizontal picture, I doubt I’m that far off.

The Grande Scheme of Thyngs

Exactly how close this set-up is to true macro doesn’t really matter, though it is fun to figure out (and useful in comparing to other lenses). What *does* matter is the image quality. After all, using a lens designed for 35mm film might mean poor image quality, even in perfectly focused situations, on the E-510’s smaller sensor.

Fortunately, that is not the case with this lens. I consistently see images coming from this lens + tube that have more crisp detail than I typically see from my other lenses. That might be because I can take the time to do a better focusing job in tabletop photography than the AF mechanism can in hand-held shooting, but at least I know this lens is great. (I haven’t tried non-macro photos as focusing is not optimal in the E-510’s small, non-focus-equipped viewfinder)

When studying the above images, keep in mind that the full-size version you get when you click them has not been sharpened at all, and it came from RAW files. As cameras always apply some in-camera sharpening algorithm to their processed JPEGs, it’d be unfair to compare them to other photos without taking this into account. As an aside, the E-510 lets you control in-camera JPEG sharpening, which is a unique and awesome, if not well-documented, feature.

Also keep in mind that the poor image quality at the edges of those photos is not a lens issue. In macro-photography we deal with very, very shallow depths of field - hence the need to focus by moving the camera with a shallow-threaded screw. I took a picture of a straight object, which means that the distance from the edges to the sensor is greater than in the middle, and in this case they lie outside the area of acceptable focus. If I’d bent the rul…, er, scale so that it formed an arc, I could have gotten the whole thing in focus.

(I suppose I could have composed the photos a lot better, too, more horizontal, closer to center, more attention to critical focus… but when you get right down to it, I’m lazy)

Trackback URI | Comments RSS