Will DPReview admit they need to eat some crow?

Started Sep 15, 2009 | Discussions thread
OP photo nuts Senior Member • Posts: 1,364
Yes, and that can be shown

Consider the various DXOMark plots for Nikon D3, D3X, D90 and Canon 40D at the pixel level:

(a) DR of D3X and 40D have different trends. The 40D DR plot is a curve that seemingly plateaus at low ISO while the D3X is a straight line.

It's been suggested that this difference is due to different ADC (analog to digital) ar-chitecture. The straight line DR plot is a signature of Sony's architecture which integrates onto the sensor chip an analog to digital converter in every column.

(b) Whereas the SNR 18% curve for D3X and 40D are straight lines, that for the D90 has a slight kink at ISO 6400. This is not difficult to understand 'cos ISO 6400 for D90 is fake.

Now, what really gives away the fact that Nikon cooks some of their RAW data is the DR curves for various cameras. Whereas D3X and 40D DR plots are smooth (either straight line or curve), the same plots for the D90 and D3 are full of kinks. This tells us the D90 and D3 RAW files have been cooked in-camera.

The D3X RAW data is by far the most impressive. It is uncooked and has better DR than 1Ds3 and 40D at all ISO. Interestingly enough, 1Ds3 has the same DR as 40D at the pixel level.

(c) Examine high ISO RAW (NOT jpeg) images from the D90. You'll see that the edges of objects are smeared unlike the low ISO samples. Those are simply noise reduction artifacts

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow