This is a review on compact lenses for use with Canon APS-C cameras (e.g., EOS 20D, EOS 400D Digital Rebel XTi) for general and portrait photography. I have been using a Canon EF-S 17-85mm on a Canon 20D since 2004 and have been reasonably satisfied with its performance within its f/4 to f/5.6 aperture limits. I wanted to see if newer f/2.8 aperture lenses in the 18-50mm range from Canon, Tamron and Sigma would offer substantial advantages in performance.

Lenses Tested

Canon EF-S 17-55mm f/2.8 IS USM Canon EF-S 18-55mm f/3.5-5.6 USM Canon EF-S 17-85mm f/4-5.6 IS USM Tamron SP AF 17-50mm f/2.8 XR LD Three new Tamron lens samples were tested. Sigma 18-50mm f/2.8 EX DC Macro Lens Specifications

50% MTF and Resolution Tests

Methods Used

Overall, each lens tested offered very good to excellent performance in 50% MTF and resolution tests over the 18 to 50mm focal length settings. Resolution performance was at or above 65 line pairs per mm in the target at all focal lengths with one exception. The lowest performer was the Tamron 17-50mm f/2.8. It had the lowest resolution at 50mm and low microcontrast performance at 18mm, 35mm and 50mm settings. This lower performance by the Tamron shows in the 50mm portrait test below. The Canon EF-S 17-85mm gave very good performance in its limited aperture range.

50% MTF with Canon EOS-20D 18mm Focal Length	Resolution with Canon EOS-400D XTi 18mm Focal Length
50% MTF with Canon EOS-20D 24mm Focal Length	Resolution with Canon EOS-400D XTi 24mm Focal Length
50% MTF with Canon EOS-20D 35mm Focal Length	Resolution with Canon EOS-400D XTi 35mm Focal Length
50% MTF with Canon EOS-20D 50mm Focal Length	Resolution with Canon EOS Rebel XTi 50mm Focal Length
50% MTF with Canon EOS Digital Rebel XTi 85mm Focal Length	Resolution with Canon EOS Digital Rebel XTi 85mm Focal Length

Portrait Function Test with Mannequin

Portrait Test with Mannequin Lenses were tested at 50mm for sharpness, bokeh and out of focus aperture diaphragm patterns. Lenses tested with a Canon Digital Rebel XTi: EF-S 17-55mm f/2.8 IS @ f/2.8 EF-S 18-55mm f/3.5-5.6 @ f/5.6 EF-S 17-85mm f/4-5.6 IS @ f/5.6 EF 50mm f/1.4 @ f/2.8 EF 50mm f/1.8 @ f/2.8 Tamron 17-50mm f/2.8 @ f/2.8 Sigma 18-50mm f/2.8 Macro @ f/2.8 The EF-S 17-55mm f/2.8 IS @ 50mm f/2.8 was used in other 50mm bokeh tests here. Lighting and image processing: Shot at night with tungsten modeling light bounced off angled ceiling. RAW images processed with Capture One Pro v. 3.7.6. Additional unsharp mask, curves, brightness and contrast adjustment in Photoshop. Thumbnail images on this page are full frame.

Tamron and Sigma lenses were shot from the same tripod location as the other 50mm lenses and have a wider field of view. Therefore, focal length of the Tamron and Sigma lenses at this working distance appears to be closer to 47 or 48 mm, rather than 50mm. Conclusion from Mannequin tests: One of the major reasons for me to own a f/2.8 lens aperture lens in the 18-50mm range on an APS-C camera is to use it for portrait photography at 50mm. I was looking for sharp images and good bokeh that would rival those produced by a 50mm prime lens. I printed the images produced in this mannequin test at 8x10 inch and 12x14 inch enlargements. After looking at the web jpg's and the prints, it was no contest. The images/prints produced by the EF-S 17-55mm f/2.8 IS and the primes "pop out" with sharpness and contrast compared at f/2.8 to those produced with the other zoom lenses. For other portrait application considerations, see: Portrait lens tests for canon APS-C cameras .

Performance at 17mm or 18mm with a Parking Garage Target - Assessment of image edge sharpness and linear distortion

Click here to view 1600 pixel wide images and assess image quality Click here to view reference lines on top floor to assess barrel distortion - The EF-S 17-55 shows considerably less linear distortion than the other lenses in the 17-55mm range. The 17-85 predictably shows the most barrel distortion associated with its greater zoom range. Click here to peep pixels in the far left edge of the image in a 11 x 17 inch print file - The EF-S 17-55 wins here for freedom from linear distortion and edge image quality. The most noticeable difference in print quality to me was the absence of significant barrel distortion in the image produced by the EF-S 17-55. To see differences in edge sharpness, you need to put your nose within 30cm of the 11 x 17 inch hard-copy print.

Performance at 35mm - Landscape

Click here to view 1600 pixel wide images and assess image quality All lenses at f/8 produce an 11 x 17 inch print that looks about the same on the wall. The cheap Canon EF-S 18-55mm has noticeably softer edges if you look closely at a print. The expensive Canon EF-S 17-55mm lens produces slightly sharper image edges if you look closely with your nose 30cm from the print. An EF 24-105mm L image is included at the same focal length setting and f/8.

Performance at 85mm - Landscape Detail

Click here to view 1600 pixel wide images and assess image quality at f/8 with the EF-S 17-85 Sorry, only the EF-S 17-85mm lens will get this image unless you want to get in a boat with your EF-S 17-55mm IS. Included are images produced with an EF 85mm f/1.8 at f/8 and an EF 70-300mm f/4.5-5.6 DO IS at an 85mm setting. The prime gives a slightly sharper image than that produced by the EF-S 17-85 @ 85mm. However, the 11 x 17 inch print produced with the 17-85 looks very good. The DO lens has noticeably lower sharpness and contrast that shows in the print.

Vignetting

Lenses were tested at infinity focus using a led illuminated light box with double opaque white diffusion screens. All lenses were tested at largest and smallest focal length and at 35mm. Click on lenses below to assess results.

Canon EF-S 17-55mm f/2.8 IS @ 17mm, 35mm, 55mm Canon EF-S 18-55mm f/3.5-5.6 @ 18mm, 35mm, 55mm Canon EF-S 17-85mm f/4-5.6 IS @ 17mm, 35mm, 50mm, 85mm Tamron 17-50mm f/2.8 @ 17mm, 35mm, 50mm Sigma 18-50mm f/2.8 Macro @ 18mm, 35mm, 50mm The EF-S 17-85mm f/3.5-5.6 IS and EF-S 18-55 f/3.5-5.6 show significant vignetting at all apertures to f/8. All of the f/2.8 aperture lenses showed mild vignetting that is corrected by stopping the lenses down 1 or 2 stops. The Sigma 18-50mm f/2.8 Macro showed the least amount of vignetting.

Linear Distortion: Want Less?

Check out the lines and curves for each focal length. The EF-S 17-55mm f/2.8 IS has the least distortion. The wider-zoom EF-S 17-85mm f/4-5.6 has the most.

17-18mm 24mm 35mm 50-55mm 85mm

Corner Image Sharpness and Lateral Chromatic Aberration

Corners of Resolution Test Charts are shown at widest aperture and stopped down 1 to 2 stops.

18mm f/2.8 lenses generally had soft corner images at f/2.8 and improved by stopping them down 1 or 2 stops. The Sigma had the softest corners and moderate lateral chromatic aberration. The EF-S 17-85 also had soft corners even when stopped down to f/8 35mm f/2.8 lenses generally had soft corner images at f/2.8 and improved by stopping them down 1 or 2 stops. The Sigma showed the least improvement. The EF-S 17-85 had good corner performance wide open. 50mm f/2.8 lenses generally had soft corner images at f/2.8 and improved by stopping them down 1 or 2 stops. The Tamron showed the least improvement. The EF-S 17-85 had good corner performance wide open.

Flare and Internal Relections

Lenses were tested at 50mm with a snooted monolight flash in the background exposing the back of the mannequin at f/4. Click on lenses below to assess results.

Canon EF-S 17-55mm f/2.8 IS Canon EF-S 18-55mm f/3.5-5.6 Canon EF-S 17-85mm f/4-5.6 IS Canon EF 24-70mm f/2.8L Tamron 17-50mm f/2.8 Sigma 18-50mm f/2.8 Macro All of the f/2.8 aperture APS-C lenses flared under the test conditions. The Canon EF-S 17-55mm f/2.8 IS controlled flare\ internal reflections best. The Sigma 18-50mm f/2.8 Macro had the worst flare/internal reflections.

Autofocus Accuracy Under Variable Ambient Light

Lenses were tested at 50mm under tungsten modeling light at 1 meter working distance. Each lens was focused at infinity and then refocused on the right mannequin eye and the shutter released. The test was repeated 20 times for each lens, and the percentage of in focus images were recorded. The tests were run under good modeling lighting (ISO 800, 1/100th sec @ f/2.8) and low light conditions (ISO 800, 1/40th sec @ f/2.8). When autofocus problems were identified under low light conditions. The tests were repeated.

• Canon lenses showed the best performance with an EOS 400D Digital Rebel XTi under low light.
• The Tamron lens had the worst performance under low light (and very noisy autofocus motor).

Lens Percent In Focus Good Lighting Percent In Focus Low Light Test 1 / Test 2 Canon EF-S 17-55mm f/2.8 100 100 / 100 Canon EF-S 17-85mm f/4-5.6 100 100 / 100 Canon EF-S 18-55mm f/3.5-5.6 100 100 / 100 Sigma 18-50mm f/2.8 Macro 100 95 / 85 Tamron 17-50mm f/2.8 100 82 / 85

Conclusions and Opinions

CanonEF-S 17-55mm f/2.8 IS USM
Strengths: You get what you pay for with respect to image quality with this lens. The EF-S 17-55mm f/2.8 IS had 50% MTF and resolution scores closest to L-lenses or primes across the focal lengths tested. There was significantly less distortion at wide angles with the EF-S 17-55 than with all of the other lenses tested. Edge and corner sharpness delivered by this lens in large prints was noticeably the best of the lenses tested. Center sharpness and contrast had the best "punch" of the lenses tested. Bokeh produced by the lens at 50mm f/2.8 was close to the quality of that produced by the EF 24-70mm f/2.8L at the same focal length. The 17-55 was more resistant to flare than other f/2.8 lenses tested. Image stabilization is very useful for low light photography. Canon has built a lense with L performance in a relatively light and compact package.
Potential weaknesses: It has none that are photographically significant. This is the largest and heaviest of the APS-C f/2.8 lenses tested. Although the 17-55 has performance characteristics similar to an L-lens, weight has been kept low at least in part by not making a sealed, weather and dust impervious lens.

Tamron SP AF 17-50mm f/2.8 XR LD
Strengths: It was the smallest and lightest of the APS-C f/2.8 lenses tested. I considered the size and ergonomics of the lens on an EOS 400D Digital Rebel to be ideal. This lens produces sharp images if stopped down to f/5.6 or f/8. Image quality is better than that produced with the cheap kit EF-S 18-55 f/3.5-5.6 lens. Many will find the image quality produced by the Tamron 17-50 to be acceptable for portrait photographs at f/2.8. It shows good flare resistance and produces sharp corner images when stopped down.
Weaknesses: This lens had the lowest 50% MTF and resolution performance when used at f/2.8. I did not find the image quality at 50mm f/2.8 adequate to shoot portraits for even moderate enlargement (e.g., 8x10 inches). It has poor low light autofocusing performance on a Canon camera (EOS 400D Digital Rebel XTi). The autofocus motor on this lens is very noisy.

Tamron has quality control problems. The first lens I received did not autofocus properly and was exchanged for a functional lens with the dealer. If I added my time and extra shipping expenses to the price of the lens, it yielded an unfavorable performance/cost ratio. I was concerned that I may have had two bad lens samples, so I purchased a third new lens to test. It gave comparable 50% MTF and resolution performance to the first two lenses. The Tamron lens has uniformly low performance compared to the other lenses tested at f/2.8.

Sigma 18-50mm f/2.8 EX DC Macro
Strengths: This lens generally had better center-weighted performance on 50% MTF and resolution tests than the Tamron. Autofocus was fast and quiet, and vignetting was least of the lenses tested. Although it had slightly worse flare performance than the Tamron, all of the lenses tested in this review flared less than my EF 50mm f/1.4 prime.
Weaknesses: Although the Sigma 18-50 was sharper for portrait use than the Tamron, you could see significant differences in print sharpness between it and the Canon 17-55. Sigma has built a very good lens that gives excellent performance in center weighted images. Far corner performance on the Sigma was the worst of the lenses tested. In low light, the lens underperforms in accuracy against the cheapest Canon lens used in this test.

Canon EF-S 17-85mm f/4-5.6 IS USM
Strengths: Within it's aperture limits, the EF-S 17-85 is an incredibly versatile lens that offers excellent performance in a compact, light weight package. The 17-85mm range (28-135mm full frame equivalent) is excellent for travel and general photography when you can only carry one lens. It has generally excellent performance shot wide open. Image stabilization is excellent and will save the day in low light situations. I have been very happy with the infrared photos taken with this lens on a converted 20D (see gallery).
Weaknesses: Shooting portraits without good background selection is challenging when the widest aperture in the 50-85mm range is f/5.6. The lens has greater linear distortion than the other zooms tested. This lens wouldn't be your first choice for use in architectural photography.

Canon EF-S 18-55mm f/3.5-5.6 USM
I included this ultra-cheap kit zoom in the tests assuming it would perform poorly. I was pleasantly surprised.
Strengths: It is an ultralight, very compact lens that gives good performance when stopped down to f/5.6 or f/8. It performed better in low light autofocus accuracy tests than either the f/2.8 Sigma or Tamron. Distortion isn't any worse than with the Tamron and Sigma zooms.
Weaknesses: Softness is apparent in image corner and edges in large prints. This lens is good for snapshots. You'd want to use another lens for most photographs you want to enlarge to any great extent.

Gallery of EF-S 17-85mm f/4-5.6 IS USM Images

Sigma 18-50mm f/2.8 EX DC Macro Photos from Test

Canon EF-S 17-55mm f/2.8 IS Photos from Test

Index

Posted 28 May 2007; Latest revision posted 30 August 2007

© 2007, William L. Castleman