astrolucida
Offline
Image Upload: Off
Registered: Jan 07, 2005
Total Posts: 1472
Country: Finland
|
 Re: Hands-on: Sigma 18-200 OS | |
I've been now playing for a few days with my 18-200 OS. Some findings with a 30D:
1) The OS is very efficient. In fact, due to incorrect settings, I did take an image at 200mm with an exposure time of 1/6s and it was quite sharp! However, another try did not succeed. But 1/15s to 1/20s works OK. Sigma states in the original instructions that OS helps 2-3 stops while an addendum changes this into 3-4 stops. That seems to correspond to my experiences. My unverified feeling is that proper settling time (around 1 second) is very important for the best results.
2) At 18mm and 200mm the lens is amazing, even wide open. I think Sigma has seen the statistics that show that zooms are most often used at the extreme ends. There is some softness in the corners but nothing serious. Stopping 1-2 stops gets rid of all of the softness at those focal lengths and then the corners are really sharp.
3) 50mm and 135mm are very good, even wide open, though softness is a bit larger in the corners. 80mm follows quite nearby.
4) 28mm and 35mm, however, are not good. One of the corners shows a strong halation effect (due to spherical aberration?). Stopping down helps but there is some residual left even at f8, though it is visible only at 100% (at 50% it is very hard to see it).
5) At f11 the lens is very hard to distinguish from the 17-55f2.8 IS and 70-200f2.8L IS. In the center I cannot find any difference but in the corners, with the not-so-good focal lengths (28 and 35mm), there is a slight difference. Without comparing side-by-side, it would not be noticeable.
6) At f8 the lens is as good, in the center, as the 17-55 and 70-200. However, the difference becomes more pronounced in the corners.
7) At f5.6 the lens is still very good, except for the corners at 28 and 35, where they show some halation. At the center there is no significant difference to 17-55 and 70-200.
8) At 18mm wide open, the only differences to 17-55 are very softness in the corners, somewhat more distortion and a bit more vignetting.
9) The contrast is very good and I have not noticed big differences to 17-55 and 70-200.
10) The zoom ring is somewhat stiff around 50mm. The overall build quality is higher than a typical Canon zoom but nowhere near an L lens. Not far below a Sigma EX lens, though.
11) At first I was going to return my lens due to inaccurate autofocus, between about 18 and 40mm, in dim light. However, I soon realized that this is true for the 17-55 as well. Both the 17-55 and 18-200 between 18 and 50 slightly backfocus in dim light. However, both the 70-200 and 18-200 above 50mm autofocus more accurate than manual focus. This could be due to the focusing target, which could be too small for wide angle, as I did not move the tripod when changing the focal length.
12) AF accuracy can be increased significantly by using AI Servo and keeping the focusing button pressed until the lens settles down - even with static targets. One Shot focusing makes even large mistakes in dim light.
13) At near focal distances (2 meters), the range 18-50mm corresponds in field of view to that of the 17-55 at the same focal lengths. However, from 80mm to 200mm the field decreases so that at 200mm it corresponds to 135mm of the 70-200. However, near infinity (30 meters), there is no significant difference any more.
14) The manual focusing ring turns with autofocus, making it somewhat hard to grab the lens properly. WIth learning, however, a suitable grab can be found. Manual focus is not well-damped, but not unlike that of many Canon consumer zooms.
15) Using Imatest, the lens shows sharpness below primes, but not too much (20-30% less, depending on the prime). At 28 and 35mm, wide open, the corner sharpness drops over 50% from the center, which is very visible in the images.
16) At 100% crops, no difference in CA to the Canon zooms. When autofocus misfocuses, then some CA is visible in areas of strong contrast.
17) Wide open performance in comparison with 17-55 and 70-200 stopped to the same aperture (thus putting the 18-200 at a serious disadvantage):
- 18 vs 17mm at f3.5: the center is just slightly softer, corners somewhat softer
- 28mm at f4: difficult to see any difference at center; corner very clearly softer and showing prominent halation,
- 35mm at f4.5: just a hint softer at center; corner very clearly softer and showing prominent halation,
- 50 vs 55mm at f5: very difficult to see any difference at center; the corner just a hint softer - not noticeable unless comparing side-by-side,
- 80mm at f5.6: due to the larger image scale of the 70-200, softer at the center, but just relative to the size difference; the corner shows slight halation and is clearly softer,
- 135mm at f5.6: the scale difference is even larger, so hard to judge, but the 70-200 does not seem to have any real detail in the center in spite of the scale difference; the corner much softer,
- 200mm vs 135mm at f6.3: the same image area as the 70-200 at 135mm, but actually sharper in the center; the corner much softer but still usable,
- 200mm at f6.3: the 70-200 has larger scale but not much more detail in the center; the corner a tad softer.
All in all, stands amazingly well against two of Canon's best zooms, even wide open!
18) Autofocus can be heard but is not noisy, in my opinion, especially when compared to e.g. 35f2. OS is so silent that I was not able to hear it at the camera shop - but at home it can be heard to engage and disengage. One of the most silent stabilizers there is - quite near the 17-55.
Overall verdict? At first I was ready to return the lens, due to autofocus inaccuracy, but then saw similar issues with my 17-55, under dim light. In ordinary photographic situations I have gotten simply great images with my 17-55, so I don't really see this as a problem. Using AI Servo seems to get rid of some of the random behaviour of the AF.
Then I was worried about the uneven performance at 28 and 35mm. The Photozone tests show a very steep drop at 35mm for the non-OS version as well, so this is probably a design compromise for the lens (they did not test 28mm separately). However, the extreme ends are very good, and in addition to that 50mm is, too. Rick did not have that much luck at 50mm, though, so this might be sample variation.
If I need the range 28-35, stopping to f11 produces perfect images, while f8 is very good and f5.6 can be used at a pinch. Wide open performance (28 and 35mm at f4 and f4.5) is just not acceptable in the corners, while the center is sharp.
Then I was disappointed to find out that the max range was only 135mm. But with infinity (30 meters), it is 200mm. In my case 200mm is typically used for targets far away - with this lens.
Hence, at the moment I don't find any fatal faults with the lens. I do wish, though, that 28mm and 35mm did not have that halation, though. Softness is not an issue but halation is very visible under many circumstances (but not much at 50%). But as 18, 50 and 135-200mm will most probably be the most used focal lengths for me, I am very much tempted to keep the lens.
Considering the price of the lens, it is not far from amazing. The performance, at its best, is a couple of notches above what I would expect from a megazoom. At its worst (28/35mm corners, wide open), it just as I would expect from an inexpensive 3rd party megazoom. OS is clearly better than with the 80-400 OS (2 stops in my experience) while the 18-200 seems to really produce the 3-4 stops, especially at the tele end.
What is it? A 11x megazoom with usable to excellent IQ, depending on the focal length, aperture and location on the image. The OS is definitely very good. Build quality is better than I would expect. A petal-type lens hood is included but case is not. A non-rotating front element, making use of a polarizing filter easy.
What is it not? No weatherproofing, no full-time manual focusing, no fixed aperture when zooming, not parfocal when zooming. Not very fast (f3.5 to f6.3).
It remains to be seen how much there is sample variation. All in all, I can recommend this lens if the features are what you want and you are willing to learn its quirks and work around them.
Well done, Sigma!
|
Hands-on: Sigma 18-200 OS
