Originally posted by JohnBee why ISO1600 is the magic number when it comes to DR measurement etc...
There's nothing special with 1600. It's just my fetish. 1600 seems to be the figure where noise starts to be visible enough to seriously care about (at least for APSC). So, 1600 marks the "high iso" regime for me. But's that's entirely arbitrary. Except that it also sits in the exact middle between 100 and 25,600.
For the K-5, Gordon B Good additionally believes to have evidence that iso levels higher than 1600 are produced digitally from analog data read out at ISO 1600. That needs further verification though.
Originally posted by JohnBee I'm really not sure why our samples are not showing the NR threshold with the K-5... and trying to figure out why theory isn't meeting practicality.
I'm awaiting your samples before I can comment. If hunting for signs of NR, comparing ISO1600 at -3EV, and ISO 12,800, for an identical subject, would be ideal.
I assume it will be hard to see signs of NR with the naked eye. A lab test is more easily able to show the difference. E.g., for the K-r, the effect of NR is only +1.8 dB extra SNR (about half a stop) above ISO 1600. Very difficult to discern by pure visual inspection.
Originally posted by timh 'sports' ISO at 561. From experience, I always have the auto-ISO ranging from 100-500 which tells me that DxO and I agree on what acceptable quality is
Yes, DxO's sports score (aka low light ISO score) is a good default for the upper AUTO ISO limit. Which means that one doesn't need to care about bad image quality.
However, real sports photography would actually go into the "unsafe" area and accept a certain level of grain to make fast pace action shots feasible. As shown by the K-5 samples (assuming it will have a sports score of about 1000), the K-5 has no problems of producing glossy magazine action shots at, e.g., ISO 6400 (with some post processing).