Much has been written in the past few days about the Canon 5D MkII’s so-called black dot problem, in which pixels to the right (as viewed) of extreme point source highlights appear as black dots. Canon has been asked for explanations; my feeling is that the explanation is already present in the way that the 5D MkII handles the necessary process of sharpening.
5D MkII files I have shot (around 500, I’m not a prolific shooter, before the test camera went back) display a far more aggressive edge sharpening policy than any other DSLR raw file I’ve seen. Combined with a very steep midtone curve – crushing the shadows a fair bit, but not unkind to highlights – this produces some of the sharpest looking images around ‘out of the box’.
But, if that visual acuity is to exist on moderate contrast contours and transitions (dark/light boundaries – the basis for all sharpening) it may be extreme on boundaries between dark midtone and small blown highlight pixels. The directional quality of the black dot problem points to a sharpening artefact, or more accurately an edge enhancement artefact – not to be confused with post-capture or JPEG process sharpening.
If so a firmware fix is possible. It may, in the process, make the 5D MkII images from raw appear a touch softer than they do right now. These ‘black dots’ are certainly not dead pixels, and don’t have much to do with the actual sensels on the CMOS. They are created after the image is read out from the silicon.
I may be wrong, but that’s my prediction – the black dots will prove to be an artefact created by a aggressive contour sharpening policy.