This image and the next shot are direct comparisons of the Nikon D3X at ISO 1600, in-camera JPEG, with the Sony Alpha 900 at exactly the same settings. Though the Nikon 24-120mm VR zoom reported 95mm focal length, the framing was as close as possible to the Sony set to 105mm on a 1993 vintage Minolta 28-105mm RS zoom. Both lenses were set to f8; I remained glued to the floor (feet never moved!), and hand held exposures were made at 1/60th at f8, set manually, with SSS and VR stabilisation enabled respectively. Image settings were default style, sRGB, but with High ISO NR disabled in both cases, and all other special parameters disabled - DRO, D-Lighting, Shading Correction, etc. These two shots are as plain vanilla as possible, no post processing, no NR etc.
Three exposures were made with each camera, taking care to be as stable as possible for hand-held shots by two 100W modelling lights in softboxes, using tungsten light balance. The sharpest exposure was selected for each camera. This give a slightly unfair result, as two Nikon exposures were not sharp at all but all three Sony exposures were identically sharp (the 24-120mm, at this distance, seems to be a poor second optically to the less ambitious older Minolta glass).
Both files have been resaved at JPEG level 10. Clearly, the Nikon high ISO 'grain sructure' is very different from Sony's larger noise pattern with higher chroma noise, but this is by no means the whole story as examination of the images will show. Also, the Nikon appears to 1/3-1/2 of a stop more sensitive at the same nominal ISO setting; either Sony is a true 1000-1250, or Nikon is a true 2000-2500. The tone curves are different and this may be partly why the Sony image is darker overall.
My own conclusion is that the D3X does not necessarily incorporate any miracle image processing when compared to the Alpha 900, even basing this on in-camera JPEGs; the approaches are different in the way they have always been different. In the end there is little to choose between the two, and the performance of a particular lens or the accuracy of focus for each individual shot could tip the balance either way.