>>4228434FYI, the G9 exceeds the ideal line, but pretty much all full frame cameras except the K-1 II, which has a godawful noise reduction algorithm forced ON over ISO 400 (typical pentax incompetence), struggle to approach the ideal line for FF and most are closer to the ideal line for APS-C. Most APS-C cameras are closer to the ideal line for 4/3 unless they label ISO a full stop darker (basically just fuji). 4/3 cameras started out following this trend, approximating the ideal for the next crop factor down, and then something changed.
This is what cheating on your spec sheets looks like. The manufacturing and marketing equivalent of sneaking a calculator into math class.
There is something going on. Panasonic and olympus are not sitting on sensor tech from 2050. They are combining mislabeled ISOs with forced noise reduction that p2p and DXO arent able to distinguish from natural noise (the squiggly appearance of the line does imply it however - less cooked raws have straighter trendlines) - meaning the lower requirements for processing the smaller sensors output have allowed room for an advanced noise reduction algorithm of the kind usually seen in dedicated editors. Considering the methods for detecting on camera NR are published, evading them is a triviality. They are getting better at editing the files before you do, so they can tell you the camera is better, when it's really just taking away your choice to use noise reduction or not, and telling you ISO 100 is actually ISO 200.