Sony Cyber-shot DSC-RX1 Review
The actual sensitivity of each indicated ISO is measured using the same shots as are used to measure ISO noise levels, we simply compare the exposure for each shot to the metered light level (using a calibrated Sekonic L-358), middle gray matched. We estimate the accuracy of these results to be +/- 1/6 EV (the margin of error given in the ISO specifications). In our tests we found that measured ISOs from the RX1 match the marked ISOs within 1/6 stop accuracy, meaning ISO 100 indicated = ISO 100 measured.
Noise and Noise Reduction (JPEG)
The RX1's JPEG produces similar amounts of noise to the recent SLT-A99 but with increasing divergence at high ISOs. Without the SLT's semi-transparent mirror, the RX1's sensor is always receiving rather more light than the A99's, so is able to produce cleaner JPEGs (or, at least, not have to apply as much noise reduction to its files). Its noise reduction is a little more aggressive than the Nikon D600's, which explains how it appears to post better results in this test.
The RX1 only offers three noise reduction settings - off, low and normal. The 'off' setting still applies a fair dose of chroma noise reduction but leaves luminance noise relatively untouched, so you get a gritty but still sensibly-colored image.
The Low and Normal settings are only very subtlety different in this test - possibly because it applies different amounts of noise reduction to different areas of the image (in an attempt to distinguish between smooth areas, detail regions and edges). There's a possibility both settings apply the same amount of noise reduction to the areas our test samples but would differ with other subjects. That said, we've seen little difference between the two in our real-world shooting.
Adobe Camera Raw noise (ACR 7.3, noise reduction set to zero)
The RX1 gives exactly the performance you'd expect - a touch ahead of the A99 and on a par with the D600. Although it may initially seem odd that the X100 seems to produce a similar result, it's worth remembering that this is a pixel-level test. Because the X100 has a pixel count that's lower in similar proportion to how much smaller its sensor is, you'd expect it to give a similar result to the RX1.