GEAR: Canon EF 28mm f/1.8 & Sigma 30mm f/1.4 compared
Tutorials and Reviews
Registered Users, Retired Mod Posts: 138
[imgr]http://dgrin.smugmug.com/photos/60714323_dmJyC-S.jpg[/imgr]Canon 28mm f/1.8 & Sigma 30mm f/1.4 compared
Review by ubergeek
Introduction: In February 2005, Sigma announced an intriguing lens--a 30mm f/1.4 "DC" format--"DC" referring to a reduced image circle suitable for digital cameras with APS-C-sized sensors. On such cameras, the lens offers a "normal" field of view approximating that of a 50mm lens on a "full frame" 35mm camera. SLR camera giants Canon and Nikon so far seem to have ignored the "50mm equivalent" for crop-factor cameras (although in their defense, 28mm and 35mm lenses are at least in the ballpark).
As an owner of a Canon 28/1.8 and an addict of big apertures, I found myself wondering how the 30 would stack up, particularly in terms of sharpness. The 28 is the lens I reach for most often (it serves as a "normal" lens on my Digital Rebel XT), and while I'm quite fond of it, I was tempted by the extra 2/3 stop offered by the Sigma. There is of course Canon's 35mm f/1.4L, but its price is prohibitive for many of us hobbyists, and at a 56mm field of view when used with the XT, it's a little on the long side. So I set out to determine whether Sigma could dethrone my current favorite, and acquired a copy of the 30 to put the two to the test.
The Test: I'm not a professional tester of photographic equipment, nor am I the most demanding user. But I wanted to devise a meaningful test, so I was reasonably careful in the test methodology: I mounted the camera (the 8MP Digital Rebel XT) on a tripod, enabled mirror lock-up and used a remote shutter release. At each aperture setting, I took three shots with each lens and picked the best of the three to mitigate any camera shake that may have occurred. Also, rather than take pictures of a newspaper or some other subject that nobody photographs in real life, I tried to choose a photographically interesting subject.
Camera Settings: All shots were taken at the camera's full 8MP resolution, ISO 100, "sunny" white balance, aperture priority, in RAW format, using light from a nearby window. (Because the light wasn't entirely controlled, there are minor exposure variations from shot to shot.) The apertures chosen were f/1.4 (the Sigma's maximum), f/1.8 (the Canon's maximum), f/2 through f/16 in one-stop increments (f/16 being the Sigma's minimum), and f/22 (the Canon's minimum).
Focusing: I used the center autofocus point, letting the camera autofocus with each shot; incidentally, there were no focus misses with either lens. It has been argued that this methodology "tests" the AF system rather than absolute lens sharpness. While I completely agree, keep in mind that the purpose of this test is to ascertain the performance of these lenses in typical use--autofocus tolerances and all.
Post-Processing: The shots were post-processed using Capture One LE using modest sharpening and noise suppression parameters:
Test Scene: The subject I chose is a pair of objects from my home, as shown below. The objects are arranged such that the areas of detail in the center and corner fall (within my ability to estimate) in the same focal plane. The accompanying (downsized) image happens to be from the 28mm at f/11, but all shots with the 28 were taken from the same position. The 30mm shots were taken from a few inches back to achieve the same framing. The center autofocus point falls roughly on the "III" on the clock face.
100% Crops: Following are 100% crops of the resulting photos, taken from the center and (near) a corner of the image. Images produced by the 30mm are always on the left, with the 28mm images always on the right. The images appearing within these pages are scaled down; click each image to see the 100% crop (make sure also that you select "Original" size).
Center at f/1.4. Obviously only the Sigma opens up this far.
Center at f/1.8.
Center at f/2.0.
Center at f/2.8.
Center at f/4.
Center at f/5.6.
Center at f/8.
Center at f/11.
Center at f/16.
Center at f/22. (The Sigma's minimum aperture is f/16.)
As seen in the full scene, the "corner" isn't literally in the corner of the image, but as far toward the corner as I would put anything important in a scene.
Shown is the corner at f/1.4. (Again, the Canon's maximum aperture is f/1.8.)
Corner at f/1.8.
Corner at f/2.0.
Corner at f/2.8
Corner at f/4.0
Corner at f/5.6
Corner at f/8
Corner at f/11
Corner at f/16
Corner at f/22. (Again, the Sigma stops down only to f/16.)
Image Analysis: Again, I'm not a professional tester of photographic equipment, just an enthusiast. However, I will offer my observations of the two lenses.
Exposure: When initially viewing the samples side-by-side, I saw that the Rebel XT appeared to consistently overexpose by about 1/3 stop with the Sigma (I consider the exposure with the 28mm to be about right). The camera selected the same shutter speed on both lenses for a given aperture, but the resulting image with the Sigma was slightly overexposed. Since this phenomenon is more or less consistent across the aperture range, I don't consider this a serious issue--I would merely set the camera for a little bit of negative exposure compensation when shooting with this lens. In fact, the 2/3-stop aperture advantage plus the 1/3-stop exposure difference makes the 30 effectively a full stop faster than the 28. This means that either the 30 is actually faster than f/1.4, or the 28 is actually slower than f/1.8, or a little of both, but I'm not sure which.
In order to facilitate a more direct sharpness comparison, I "normalized" the exposures of the shots taken with the 30mm to match those of the 28mm as closely as possible. Unfortunately, because I didn't see the difference until after the fact, the exposure compensation had to be done in post-processing. However, since the compensation required was at most 0.4 stops, I don't feel that the test results were noticeably affected (other than that the blown highlight areas may be slightly larger with the Sigma, but that shouldn't affect one's evaluation of sharpness).
Sharpness: Because autofocus and a three-dimensional test subject were used, this isn't a true sharpness test. (The manufacturers publish MTF charts, and I see no need--nor do I have the capaibility--to reproduce that.) However, it's fair to draw some conclusions about the lenses' relative performance.
Both lenses are pretty darned sharp wide open, considering that they are, well, wide open. For center sharpness, the Canon may have a slight edge at f/1.8, but either lens is admirably resolving details by f/2.8. Corner sharpness is a different story, with the 28mm turning in generally superior performance. Both lenses exhibit fairly soft corners wide open (although more than adequate, especially considering these lenses' primary use as available-light shooters). By f/2.8 the Canon is looking noticeably better than the Sigma, although the latter catches up at f/8. Keep in mind that the 28mm's image circle fills an entire 35mm frame, so a "cropped" camera such as the Rebel XT is using the "sweet spot" of the lens. The 30's APS-C-sized image circle, on the other hand, means it has to work harder in the corners, but it turns in a decent performance, considering.
Non-Image Considerations: Besides the quality of the images produced by these two lenses, there are other comparisons to be made, such as size and weight, expense, and features that contribute to (or detract from) overall usability.
Size: The Sigma is a few millimeters larger than the Canon, which is likely to go unnoticed. However, even though the Sigma produces a smaller image circle, its f/1.4 aperture no doubt requires a bit of glass. So it gains about 100g, or 4 ounces, over the Canon. By way of comparison, Canon's 50mm/1.4 is ever so slightly smaller and lighter than the 28/1.8 (although it looks larger with its hood attached). Given the Sigma's similarity to Canon's 50 (equivalent aperture and roughly equal focal length in 35mm-equivalent) and its smaller image circle, it seems that it should be commensurately smaller and lighter, so its relative size is something of a mystery. However, in practice its additional bulk isn't noticeable.
Cost: On the used market (at least in the US), both the 28/1.8 and the 30/1.4 can be found in the $325 range if one is patient. The Canon buyer will, however, also want to purchase the EW-63 II hood, making the effective cost slightly higher (whereas a hood is included with the Sigma).
The market for new lenses also demonstrates relative parity: Again in the US, either lens can be found for about $400. Again, the Canon buyer will also want a hood, making the true cost of the Canon a little bit higher.
Usability: The Sigma is every bit the Canon's equal in terms of build quality and autofocus performance. Sigma's HSM, like Canon's USM, is quiet and fast, and both lenses focus accurately on my Rebel XT. Also like USM, HSM enables "full time manual focus," allowing the photographer to manually tweak focus after the autofocus system has selected focus. Neither lens' autofocusing causes the focusing ring to rotate, which probably isn't a big deal but I certainly prefer it that way.
It's possible that the 30/1.4's larger aperture may enable it to autofocus in some extremely low lighting conditions where the 28/1.8 would have trouble. In practice, in conditions where there's enough light for me to see, my 28 is able to focus, so this advantage may be only theoretical. By the way, for all this talk about maximum apertures, it should be noted that the 30's minimum aperture is f/16, while the 28 offers another stop down to f/22.
One last observation is that the Canon provides a depth-of-field scale (with an infrared focus mark) on the focus scale, whereas the Sigma doesn't. While many photographers would probably never notice these omissions, a depth-of-field scale is occasionally useful.
Conclusion: So which one would I buy? That's easy--I bought them both! :-) However, I bought the Sigma with the intention of keeping the better lens, so the real question is, which one would I keep? Both are great lenses, but it comes down to this:
![http://dgrin.smugmug.com/photos/60714323_dmJyC-S.jpg[/imgr]](http://dgrin.smugmug.com/photos/60714323_dmJyC-S.jpg[/imgr]){kind=link}
Review by ubergeek
Introduction: In February 2005, Sigma announced an intriguing lens--a 30mm f/1.4 "DC" format--"DC" referring to a reduced image circle suitable for digital cameras with APS-C-sized sensors. On such cameras, the lens offers a "normal" field of view approximating that of a 50mm lens on a "full frame" 35mm camera. SLR camera giants Canon and Nikon so far seem to have ignored the "50mm equivalent" for crop-factor cameras (although in their defense, 28mm and 35mm lenses are at least in the ballpark).
As an owner of a Canon 28/1.8 and an addict of big apertures, I found myself wondering how the 30 would stack up, particularly in terms of sharpness. The 28 is the lens I reach for most often (it serves as a "normal" lens on my Digital Rebel XT), and while I'm quite fond of it, I was tempted by the extra 2/3 stop offered by the Sigma. There is of course Canon's 35mm f/1.4L, but its price is prohibitive for many of us hobbyists, and at a 56mm field of view when used with the XT, it's a little on the long side. So I set out to determine whether Sigma could dethrone my current favorite, and acquired a copy of the 30 to put the two to the test.
The Test: I'm not a professional tester of photographic equipment, nor am I the most demanding user. But I wanted to devise a meaningful test, so I was reasonably careful in the test methodology: I mounted the camera (the 8MP Digital Rebel XT) on a tripod, enabled mirror lock-up and used a remote shutter release. At each aperture setting, I took three shots with each lens and picked the best of the three to mitigate any camera shake that may have occurred. Also, rather than take pictures of a newspaper or some other subject that nobody photographs in real life, I tried to choose a photographically interesting subject.
Camera Settings: All shots were taken at the camera's full 8MP resolution, ISO 100, "sunny" white balance, aperture priority, in RAW format, using light from a nearby window. (Because the light wasn't entirely controlled, there are minor exposure variations from shot to shot.) The apertures chosen were f/1.4 (the Sigma's maximum), f/1.8 (the Canon's maximum), f/2 through f/16 in one-stop increments (f/16 being the Sigma's minimum), and f/22 (the Canon's minimum).
Focusing: I used the center autofocus point, letting the camera autofocus with each shot; incidentally, there were no focus misses with either lens. It has been argued that this methodology "tests" the AF system rather than absolute lens sharpness. While I completely agree, keep in mind that the purpose of this test is to ascertain the performance of these lenses in typical use--autofocus tolerances and all.
Post-Processing: The shots were post-processed using Capture One LE using modest sharpening and noise suppression parameters:
- Method: Standard look
- Amount: 50
- Threshold: 3
- Noise suppression: 2 (Low=1, High=4)
- Color Noise suppression: 15
- Banding suppression: 2 (Low=1, High=4)
100% Crops: Following are 100% crops of the resulting photos, taken from the center and (near) a corner of the image. Images produced by the 30mm are always on the left, with the 28mm images always on the right. The images appearing within these pages are scaled down; click each image to see the 100% crop (make sure also that you select "Original" size).
Center at f/1.4. Obviously only the Sigma opens up this far.
Center at f/1.8.
Center at f/2.0.
Center at f/2.8.
Center at f/4.
Center at f/5.6.
Center at f/8.
Center at f/11.
Center at f/16.
Center at f/22. (The Sigma's minimum aperture is f/16.)
Shown is the corner at f/1.4. (Again, the Canon's maximum aperture is f/1.8.)
Corner at f/1.8.
Corner at f/2.0.
Corner at f/2.8
Corner at f/4.0
Corner at f/5.6
Corner at f/8
Corner at f/11
Corner at f/16
Corner at f/22. (Again, the Sigma stops down only to f/16.)
Exposure: When initially viewing the samples side-by-side, I saw that the Rebel XT appeared to consistently overexpose by about 1/3 stop with the Sigma (I consider the exposure with the 28mm to be about right). The camera selected the same shutter speed on both lenses for a given aperture, but the resulting image with the Sigma was slightly overexposed. Since this phenomenon is more or less consistent across the aperture range, I don't consider this a serious issue--I would merely set the camera for a little bit of negative exposure compensation when shooting with this lens. In fact, the 2/3-stop aperture advantage plus the 1/3-stop exposure difference makes the 30 effectively a full stop faster than the 28. This means that either the 30 is actually faster than f/1.4, or the 28 is actually slower than f/1.8, or a little of both, but I'm not sure which.
In order to facilitate a more direct sharpness comparison, I "normalized" the exposures of the shots taken with the 30mm to match those of the 28mm as closely as possible. Unfortunately, because I didn't see the difference until after the fact, the exposure compensation had to be done in post-processing. However, since the compensation required was at most 0.4 stops, I don't feel that the test results were noticeably affected (other than that the blown highlight areas may be slightly larger with the Sigma, but that shouldn't affect one's evaluation of sharpness).
Sharpness: Because autofocus and a three-dimensional test subject were used, this isn't a true sharpness test. (The manufacturers publish MTF charts, and I see no need--nor do I have the capaibility--to reproduce that.) However, it's fair to draw some conclusions about the lenses' relative performance.
Both lenses are pretty darned sharp wide open, considering that they are, well, wide open. For center sharpness, the Canon may have a slight edge at f/1.8, but either lens is admirably resolving details by f/2.8. Corner sharpness is a different story, with the 28mm turning in generally superior performance. Both lenses exhibit fairly soft corners wide open (although more than adequate, especially considering these lenses' primary use as available-light shooters). By f/2.8 the Canon is looking noticeably better than the Sigma, although the latter catches up at f/8. Keep in mind that the 28mm's image circle fills an entire 35mm frame, so a "cropped" camera such as the Rebel XT is using the "sweet spot" of the lens. The 30's APS-C-sized image circle, on the other hand, means it has to work harder in the corners, but it turns in a decent performance, considering.
Non-Image Considerations: Besides the quality of the images produced by these two lenses, there are other comparisons to be made, such as size and weight, expense, and features that contribute to (or detract from) overall usability.
Size: The Sigma is a few millimeters larger than the Canon, which is likely to go unnoticed. However, even though the Sigma produces a smaller image circle, its f/1.4 aperture no doubt requires a bit of glass. So it gains about 100g, or 4 ounces, over the Canon. By way of comparison, Canon's 50mm/1.4 is ever so slightly smaller and lighter than the 28/1.8 (although it looks larger with its hood attached). Given the Sigma's similarity to Canon's 50 (equivalent aperture and roughly equal focal length in 35mm-equivalent) and its smaller image circle, it seems that it should be commensurately smaller and lighter, so its relative size is something of a mystery. However, in practice its additional bulk isn't noticeable.
Cost: On the used market (at least in the US), both the 28/1.8 and the 30/1.4 can be found in the $325 range if one is patient. The Canon buyer will, however, also want to purchase the EW-63 II hood, making the effective cost slightly higher (whereas a hood is included with the Sigma).
The market for new lenses also demonstrates relative parity: Again in the US, either lens can be found for about $400. Again, the Canon buyer will also want a hood, making the true cost of the Canon a little bit higher.
Usability: The Sigma is every bit the Canon's equal in terms of build quality and autofocus performance. Sigma's HSM, like Canon's USM, is quiet and fast, and both lenses focus accurately on my Rebel XT. Also like USM, HSM enables "full time manual focus," allowing the photographer to manually tweak focus after the autofocus system has selected focus. Neither lens' autofocusing causes the focusing ring to rotate, which probably isn't a big deal but I certainly prefer it that way.
It's possible that the 30/1.4's larger aperture may enable it to autofocus in some extremely low lighting conditions where the 28/1.8 would have trouble. In practice, in conditions where there's enough light for me to see, my 28 is able to focus, so this advantage may be only theoretical. By the way, for all this talk about maximum apertures, it should be noted that the 30's minimum aperture is f/16, while the 28 offers another stop down to f/22.
One last observation is that the Canon provides a depth-of-field scale (with an infrared focus mark) on the focus scale, whereas the Sigma doesn't. While many photographers would probably never notice these omissions, a depth-of-field scale is occasionally useful.
Conclusion: So which one would I buy? That's easy--I bought them both! :-) However, I bought the Sigma with the intention of keeping the better lens, so the real question is, which one would I keep? Both are great lenses, but it comes down to this:
- Being the "aperture junkie" that I am, I find it hard to ignore the Sigma's 2/3-stop advantage.
- While the Canon may be a tiny bit sharper in the center, the difference isn't enough (to me) to offset the Sigma's f-stop advantage.
- For a "normal" lens, I'm not overly concerned about having maximum corner sharpness, as I'll be shooting wide open most of the time. At such large apertures, objects other than the subject(s) will likely be out of focus due to the shallow depth of field anyway.
- The idea of reduced-image-circle lenses doesn't bother me. I'll probably be using a 1.6x crop-factor camera for some time, and even if/when I do switch to "full frame," I'll start using my 50mm/1.4 as a "normal" lens instead.
- If I really need exposures slower than f/16, I'll just get myself a neutral density filter. :-)
0