Okay, so I did a TON of research before buying my newest lens, a Sigma 30mm f/1.4 "fixed" or "prime" (meaning it doesn't "zoom", its always at 30mm). But, in all my research I read over and over again that this lens was "L series" sharp in the center, but it's corner and edge performance was poor. The unanswered question was -where's the "center" VS the "edge"? Where does the sharpness fall off? How big of an area will be "L series" sharp? There seemed to be no answer to that question, but I bought it anyway based on a lot of SQF and MTF data as well as lens tests others have done (all archived and/or linked to on this page).
Now that I've had the lens for a couple weeks, I decided to run my own lens sharpness tests. This was my first attempt, so I wasted a whole day doing tests by taking pictures of a newspaper ad taped to the wall. After taking shots with EVERY LENS I OWN (and at several focal lengths for the zoom lenses!), I converted the resulting photos (over 250 of them) from RAW to jpeg, then started comparing them. It was then that I realized that my tests were flawed. The camera was not always exactly perpendicular to the wall, so the edges on many shots were blurry -not because the lens was flawed, but because they weren't in the plane of focus! (GRR!)
So, last night I re-did the test, but only with my Sigma 30mm f/1.4 and my Canon 100mm f/2.8. This time I laid some paper with printing on it on the floor, then used a level to make sure the camera's LCD screen was level before taking the photos (straight down). I had to redo these several times to try to get the framing of the photos to be roughly the same at both focal lengths, but I finally got it right.
All photos were taken on a tripod, using a remote shutter release, with mirror lockup enabled to get the most sharpness possible from each shot. Also, 2 photos were taken at each focal length so that if camera shake was an issue it'd be obvious when the 2 shots were compared.
After the photos were taken, the two versions were compared to make sure they were the same, then they were converted to jpeg. Now, because what I wanted to see when I was shopping around was an entire, full-resolution shot, I uploaded all the shots to flickr, they are in this set.
Then, I opened all the shots at once in Gimp, found the exact center of the photo, and then expanded them all to 100% so that they'd show the same area. Next, I arranged all the windows in order and in such a way that the shots with both lenses at each focal length were together for easy comparison, and then I took a screenshot and uploaded that photo as the "center crops" photo.
Then I looked for the clearest edge for both lenses , and took a crop from as close to the middle of the bottom and right-hand edges as possible, and with them arranged the same as the were for the center crops, I took another screenshot and uploaded those photos as the "bottom edge" and the "right edge" crops. And lastly, I did the same thing for the "corner crop" photo.
All the photos were added to a set which can be seen here: http://www.flickr.com/photos/erica_marshall/sets/72157602932628392/
I'm not going to dwell on results here, because you can see them for yourself. I do feel that it's really important to put this data in perspective, though, so here are some tips for using these test shots in the way they were intended:
If you compare the two lenses, you'll see that the Canon 100mm is much sharper through to the edges and corners than the 30mm. This isn't much of a surprise and does NOT say anything bad about the Sigma 30mm f/1.4. This dramatic difference is due to the focal length -it's much easier to get clear edges and corners with a longer focal length. My suspicion is that even the Canon 35mm f/1.4L lens isn't completely sharp in the corners!
These tests show black and white text to maximize the lack of sharpness. In the real world the difference in sharpness that you see here will almost be imperceptible. The differences will never be this clear in a real photo.
This is a fast lens who's use at low-numbered apertures (which is kindof the point of having it) will yield a very shallow depth of field. When you have a shallow DOF, your subject is usually not right on the edge or in the very corner of the frame anyway, so edge and corner performance isn't really all that important on this type of lens.
Keep in mind that none of these shots were sharpened at all, they are straight RAW files, converted to jpegs.
Don't fall into the trap of obsessing over numbers. I obviously have to an extent, but you have to be able to take a step back and see the whole picture. These tests may be flawed and the results may not apply in real world situations. Use at your own $risk$.
Perhaps one day I'll try to rent a Canon 28mm f/1.8 and re-do the tests so they can be compared more directly. But, I find it interesting to see how much of a difference aperture can make in the sharpness of your photos. In fact, I think I'm going to test the rest of my lenses as well, then label each lens with it's sharpest apertures and some indicator of where the sharpest part of the frame is as well.
As close to a conclusion as I'll get for now:
Last night I was asked if I was happy with my purchase or if I regretted it. I must say that I've already gotten use out of it in low-light situations where I never would have been able to manage a picture before, so I am finding it useful. As far as sharpness goes, I think it's acceptable in the center, and I know, now, how to get the best photos from it. But, not having other similar lenses to test, I cannot say for sure if I'd be happier with another (similar) lens. -I guess time will tell as far as how well it does in real-world situations. I have been impressed with it's use, like I said, but I haven't really had a photo yet that puts it to the test as far as sharpness goes. I do intend to use this lens a lot in my home studio, though, so I'll definitely post another review whenever I get around to getting a really good real-world picture with it.