Velvia 100 + Nikon Coolscan 4000 versus Canon 5D

This comparison has been originally made to confirm or dispel the myth of decreasing sharpness of full frame sensors in the corners of an image. Since this effect is said to be most pronounced with wide angle lenses, the sharpness comparison is made with the 17-40/4 L zoom lens at its widest setting and f/11. A small part of the scanned image is masked by the slide mount. The scans were saved as jpegs, with a compression 10 on a scale 1 to 12 to limit file size, possibly loosing a teeny bit of detail. The scanned image has a small curves correction applied. The tiny black dots that you can see in the scanned images are the so-called pepper grain, which has nothing to do with film grain. These are small bumps in the base of the film emulsion picked up by the scanner. They can be eliminated by using ICE, which may affect the sharpness of the image and my goal was to avoid that. The quality of the scan is limited by the scanner and outside the center of the image also by the lens used.

The image from 5D is straight from DPP.

Both the film image and the 5D image were focused manually using a viewfinder magnifier. No re-sampling has been applied. Draw you own conclusions, copy the images at will, but cite the source.

Velvia 100 straight from the scanner
Velvia 100 + a bit of USM
Canon 5D, default settings, sharpness 0, jpeg 9
Canon 5D, default settings, sharpness 3, jpeg 9
Canon 5D, default settings, sharpness 3, jpeg 10

Another comparison with the 17-35/2.8L at 17mm regarding vignetting in the corners is coming up next. The conclusion for this test will be entirely mine and objective, based on measurements.

Last updated: February 1, 2006
Please send any comments to

or, if you prefer, sign an entry in the guest book.