There are geeks and then there are photo geeks. In the old days, geeks worked in carnivals and were oddly entertaining folks who swallowed swords, hammered spikes into their nostrils, and decapitated chickens and snakes without benefit of a cleaver. The photo geek, by contrast, is not nearly so riveting. In fact, photo geeks are downright dull. They photograph things like test charts and brick walls, and talk about spherical aberrations and aperture diffraction rather than composition, light, and shadow.
In general, I tend to avoid partaking in the nerdier aspects of photo geekiness. Although I enjoy reading about lens coma and apochromatic lenses as much as the next (obsessive) photographer, I scarcely resort to photographing graphs or grids for the purpose of quantifying camera or lens performance. Instead, I quantify performance by how capably the gear enables me to capture real-world subjects. I choose these subjects for a multitude of reasons, but rarely for the purpose of measuring or calculating subtle optical flaws.
I only consent to such behavior when these real-world photos display optical aberrations that are either problems to rectify or mysteries to solve. In these instances, I put on the white coat, descend into my metaphorical laboratory, dust off the old slide rule and begin photographing charts, graphs, and dreadfully insipid objects… which brings us to the purpose of today’s article.
Ever since I began sharing M-mount lenses between the Leica M8 and the Panasonic DMC-G1, I’ve been consciously aware that they perform quite differently on the two cameras. Specifically, photos on the M8 tend to be consistently sharp across the frame, whereas photos from the G1 (when fronted with an adapted M-lens) tend to get rather soft in the corners. Because of the way I use these lenses on the G1, this hasn’t concerned me much. I’ll pop an M-mount lens on a G1 for one of two reasons: either I need faster glass (to compensate for the G1’s rather poor high-ISO performance), or I’m taking advantage of the G1’s 2x crop-factor to ‘convert’ a lens into a mid-range telephoto. In both cases, I tend to shoot subjects that are somewhat centrally-located, and the reduced depth-of-field more than masks any softness in the corners.
Quite frankly, if it weren’t for that darned internet, I would probably never have fired up the Tesla coil, filled the Erlenmeyer flasks with brightly colored liquids, nor had my assistant fetch me a fresh corpse. I’d happily continue to mount M-lenses on the G1 — mindful of the subject and whether corner softness would negatively impact the photo — never bothering to ‘quantify’ the softness in a lab. However, because of my various blogs about the M8 and the G1, many readers have written me to ask specifically about the differences between these cameras when using M-mount lenses.
Sean Reid was, to my knowledge, the first to publish the fact that the G1/M-lens combination produced a rather fuzzy periphery. His tests, conducted with a 28mm Summicron, confirmed what I saw in my own images. Crazy as it may seem, I’ve never actually mounted my 28 cron on the G1, but I have mounted both my 35mm and 50mm lenses — and they do display soft edges. But look at the two generic “street portraits” below. Both were taken with my 1991 v5 50mm Summicron lens. For the woman’s photo, the lens was mounted on the M8. For the man’s photo, it was mounted on the G1.
Both cameras did exactly what I asked them to do, so using real-world images to contrast and compare the two cameras is somewhat futile. So I descended into the laboratory to devise a more controlled test that would tell my readers what they wanted to know.
The problem with performing a direct comparison is that the M8 and G1 have different crop factors. A 50mm lens mounted on an M8 gives a field of view roughly equivalent to a 66mm lens on a full-frame 35mm camera. That same lens, mounted on a G1, gives a field-of-view roughly equivalent to a 100mm lens. Compounding this difference is the fact that the two cameras have different pixel counts. So, before I could even begin to compare the two cameras directly, I had to partake in some mathletic exercises.
I positioned the M8 90cm away from a board on which I glued some coins. At that distance, a 50mm lens mounted on a camera with a crop factor of 1.33 yields a 48.72cm horizontal field-of-view. Photos from my M8 are 3916 pixels wide so, for this test, my M8 would capture 80.38 pixels/cm.
I needed to insure that, when this same lens is mounted on a G1, the coins would appear to be the same size. Since the M8 test yielded 80.38 horizontal pixels/cm, I wanted the same from the G1. Its photos are 4000 pixels wide, which meant I’d need to capture a 49.76cm horizontal field of view. Working backward and factoring in the G1’s 2.00 crop factor, I calculated that the G1 would need to be 139cm from the target in order to capture an image that was dimensionally equivalent to the M8.
Below are thumbnail views of the uncropped test photos. As you can see, the field of view captured by each camera is nearly identical. On the left is the M8. On the right, the G1:
For each image, I focused manually on the Loonie in the center. Already I’ve learned something: I had always assumed it was easier to focus the G1 using its magnified focus assist feature than it was to focus the M8. But, as you can see in the following comparisons, I actually obtained slightly better center focus with the M8. I ran each test twice, and these were the best f/2 center focus results from each camera. I then tried to get sharper center performance from the G1, but could not. I can only theorize that the M8’s sensor (with its thin glass coating and missing anti-alias filter) is simply able to resolve analog, unprocessed images better than the G1. And you can’t get more “analog” than an M-mount lens.
More caveats: I shot the M8 at ISO 320 and the G1 at ISO 400. You can see that, at ISO 400, the G1 is much noisier than the M8. This doesn’t really have any effect on what we’re trying to learn, but it does account for one visible difference between G1 and M8 captures. To minimize camera shake, I used a release cable on the M8 and the timer on the G1. Light came from a big, filtered window directly to the left of the test target — not the most ideal lighting, but identical for the two cameras. Finally, because each camera renders colors a little differently, I desaturated these images so that color variations wouldn’t affect one’s perception of sharpness. I used the same default Lightroom (Camera Raw) settings on all captures.
Below are the results at f/2. Images from the M8 are on the left, and images from the G1 are on the right. Because the lens behavior was uniformly concentric — affecting all corners equally, I haven’t bothered to show crops from the other corners — just the top left:
Below are the results at f/2.8. Already I’m seeing quantifiable evidence of why I really like using this lens on the M8 at f/2.8:
And here are the results at f/4. Note that there seems to be a slight bit of focus shift happening here. Center crops at f/4 aren’t quite as sharp as they were at f/2.8, whereas the corners are a tiny bit sharper:
Here we see the results at f/5.6. Again mild focus shift is apparent. On the M8, the corner is decidedly sharper than the center. On the G1, which suffers from soft corners, the center softness and edge softness are now nearly identical.
At f/8, depth of field begins to obscure the focus shift differentials. However, we can still see that the M8 corner image is slightly sharper than the center, whereas the G1’s corner image is about equally as soft as its center. For the G1 tests, I focused the 50mm lens only once at f/2 — this was to insure any focus shift issues that existed on the M8 would be duplicated on the G1. In reality, one is most likely to focus a G1-mounted lens at the selected aperture (unless it’s too dark).
Obviously, when my 50mm Summicron is mounted on a DMC-G1, the corners are softened across all apertures. At f/2 – f/4 this is extremely apparent. On this test it isn’t as apparent at f/5.6 or f/8… until you consider that this lens has mild focus shift issues and, as such, one would expect the corners to be sharper than the center at those apertures.
What’s my conclusion? The plain and simple truth is that the M8’s sensor was designed for M-mount lenses; the Micro Four-Thirds (MFT) sensor was not. MFT cannot do justice to M-class glass. MFT really comes into its own when you use dedicated MFT lenses, since their aberrations are automatically corrected within the camera’s software. The snag, of course, is the paucity of MFT lens offerings. Personally, I’ve only used the G1’s 14-45 MFT kit lens, but I find its renderings too clinical for many of my images. It’s like the visual equivalent of Autotune™, which singers use to correct pitch errors. The problem with Autotune™ (to my ears) is that removing all of a singer’s pitch variations also removes all of the singer’s soul. I’d rather have soul than perfection, so I often choose to mount M-lenses on my G1. Your needs and aesthetics may differ. If edge-to-edge sharpness is important to your photography, you probably won’t be happy buying M-lenses to mount on your MFT camera. Also, if you already own a collection of M-lenses and are dying to use them on a digital body, you’ll be much happier spending a few extra bucks on a used Leica M8 instead of a new MFT body. If, like me, you frequently carry around both an M8 and a G1, then its hard to beat the convenience of having two “looks” for the price of one lens.
©2009 grEGORy simpson
If you find these photos enjoyable or the articles beneficial, please consider making a DONATION to this site’s continuing evolution. As you’ve likely realized, ULTRAsomething is not an aggregator site — serious time and effort go into developing the original content contained within these virtual walls.
Hi
well perhaps its because of the incident angle of the summilux? If you try something with a longer flange distance it might be better. I found that FD lenses and OM lenses work nicely on the G1 … better than the native lenses if you ask me.
I’ve put up a lot on my blog about this (listed above)
Egor,
I have to wonder if you were to pull the anti-aliasing filter from the G1 whether you would see that it is at least a partial cause of diffusing (defocussing) the circle of confusion in a gradient as you move center to edge.
The image pases into the anti-aliasing filter at an angle causing a cascade of light within the depth of the filter. This introduces more off-axis radial scatter than the symmetrical scatter of the perpendicular image forming rays in the center of the image.
I have M and R lenses, a Liecaflex SL-Mot and M-3 + 35mm Summilux W Frogeyes. I always liked the larger mag of the M-3 finder over the M-2,M-4 etc..
I use the R lenses w/adapter on my Canon 5D with no visible image defects in 17×26″ prints. (Canon IPF5000 printer)