Originally posted by clackers Photonstophotos is Bill Claff, a Harvard professor and amateur
Bill Claff's homebrew site is
at least as unreliable as DxO. It is the same nonsense.
1) His numbers do not match visual results from developing raw images, so they are completely worthless. You'll get replies based on theory alone why figures are right and not what you see...
2) His number are based on a private and completely deliberate definition which skew results in one technical direction, making it worthless to anyone who might not follow his personal opinion. The definition is different from DxO's. The difference being that DxO's definitions includes all data and Claff privately decides to ignore some of it, because in his
opinion it is not relevant.
3) As it is pure numbers and not connected to visual experience it worthless crap again. Sample: Nikon Z7/Z6 sensors dynamic range due to major banding issues is in real life crappy for about 2 stops more, making the images no better (some might say worse) than Canon sensors. The theory numbers he presents ignore that absolutely.
It is probably worth an own thread to analyze all the unreliabilities of both Dxo and Claff.
As it stands there is NO really reliable source of truth about sensor performance out there, they all are biased and skewed in one way or another.
Anyone with real interest needs to grab raw files and compare them themselves - not believe in "numbers" which mean nothing. It is photography after all.