All shots taken with identical camera settings in RAW and straight converted with Capture 4.2.1. with only change being standardised WB (high colour fluoro - 6500K). Full size images presented, no USM applied, only saved at PS CS JPEG lvl 8. ISO 400 used.
DX Lenses: 17-55 vs 18-70 - physical size differences
Shot with 35/2 at f/2 (handheld after coffee consumption).
AF-S Lenses: 28-70 vs 18-70 - physical size differences
Wide-end distortion (the bottom edge of brown desk is meant to be straight):
17-55 at 17mm, wide aperture f/2.8.
Full size (3008x2000)
18-70 at 18mm, wide aperture f/3.5.
Full size (3008x2000)
28-70 at 28mm, wide aperture f/2.8 (standing well further back than the other two)
Full size (3008x2000)
Long end - supposed test for sharpness. Images taken with camera on support, at f/8. 1/13 and 1/15 shutter speed differences IMHO negligible. Focus using centre point, double-tapped half shutter depress before release for each photo.
I estimated 55mm on the 18-70. Exif reported 52mm so I wasn't far off. What surprised me was that to this equal angle of view, the 17-55 needed to be set at 45mm. So -
17-55 at 45mm
Full size (3008x2000)
18-70 at 52mm
Full size (3008x2000)
50mm at f/8 (incommensurable with the other two, as this is right in the sweet centre of this lens' zoom range - and taken much later than the two above attempting to equalise field of view)
Full size (3008x2000)
As I said, nothing scientific. Feel free to pixel peep at the full size images. No need to point out my testing methodology flaws, I already know... believe me.
Edit: added 28-70 into the mix after initial 17-55 versus 18-70 play session.

