The interesting thing is that a significant fraction of the Nikon sensors are actually Sony sensors. Nikon has always managed to squeeze more out of them than Sony itself.
The red Nikon box should be marked "Nikon + Sony + Toshiba" as its sensors are from three different makers in those Nikon cameras.
The "best Nikon" APS-C sensor in the D5200 is from Toshiba (see the Chipworks tear down).
The difference here seems to be "doing it all in house (with yesterdays process)" (Canon) and working with the best in the field (Nikon).
The 1/3ev lost in the SLT "partial mirror" means that for the same sensor they're always about 5 units lower. It may be the reason they're stopping making APS SLTs.
The Sony A580 scored no worse than the D7000. In fact, it showed significantly better colour response (like almost every Sony tested by DXO).
The Toshiba sensor scores good but the measurements don't reflect patterns in noise from which the sensor suffers both for stills and video. Which eats up to 2 stops from the measured DR in practise for rendering those stops unusable. Even when using good debanding tools from Topaz or Nik Dfine.
The interesting thing is that a significant fraction of the Nikon sensors are actually Sony sensors. Nikon has always managed to squeeze more out of them than Sony itself.
ReplyDeleteThe red Nikon box should be marked "Nikon + Sony + Toshiba" as its sensors are from three different makers in those Nikon cameras.
ReplyDeleteThe "best Nikon" APS-C sensor in the D5200 is from Toshiba (see the Chipworks tear down).
The difference here seems to be "doing it all in house (with yesterdays process)" (Canon) and working with the best in the field (Nikon).
The 1/3ev lost in the SLT "partial mirror" means that for the same sensor they're always about 5 units lower. It may be the reason they're stopping making APS SLTs.
The Sony A580 scored no worse than the D7000. In fact, it showed significantly better colour response (like almost every Sony tested by DXO).
ReplyDeleteThe Toshiba sensor scores good but the measurements don't reflect patterns in noise from which the sensor suffers both for stills and video. Which eats up to 2 stops from the measured DR in practise for rendering those stops unusable. Even when using good debanding tools from Topaz or Nik Dfine.
I wonder how they match up in full frame sensors?
ReplyDelete+1
Deleteand ppl still believe that Sony got worse IQ than Canon.... lol
ReplyDelete