I chose a nice challenging, contrasty scene with a lot of bright sky and deep shadows. I used same focal length (obv. not the same lens) and same exposure (also corrected for white balance). I then processed in LR4 with the express intent of pulling as much detail from shadows and highlights. I could pull a lot more out of the highlights on the 5DM3 image and ended up with an image that was HDR-like, while the D7000 image still had a blown-out sky. The 5DM3 image also just flat-out looked better, color-wise and tonally (several family members agreed.) It feels like this DR business is not adequately describing what one might see in the real world. What am I missing??
I wish I could give you an answer. Experiences similar to the one you describe is why DxO doesn't have much credibility in my book. I'm not one of those conspiracy theorists that claims they have an anti-Canon bias. My problem is the concept of trying to judge image quality through standardized lab tests. Other than tech geeks that have yet to lose their virginity, who the hell does this in real life?