Raw Dynamic Range

Leica doesn't have the best history of accessing the best-available sensors. The Kodak/TrueSense CCDs used in the M-8 and M-9 have their fans but by most objective measures, they've not been cutting-edge. The Leica M Typ 240 improved matters: moving to a CMOS sensor designed by CMOSIS. This allowed the camera to offer improved high ISO performance, as well as the ability to offer live view. However, push the Raw files and banding became apparent, which somewhat reduced the usability of its files.

Leica hasn't made any announcement about the sensor it's using in the Q. The improved video spec (1080p at up to 60 fps, compared to a 25p limit on the Typ 240) suggests either the sensor or supporting processor has changed, but the Q's would certainly continue to benefit from a sensor design that was very tolerant of light arriving at shallow angles, which Leica flagged as a benefit of the CMOSIS design.

Exposure Latitude

In our first test, we look to see how tolerant of pushing exposure the Raw files are. We've done this by exposing our scene with increasingly lower exposures at the camera's base ISO of 100, then pushed them back to the correct brightness using Adobe Camera Raw. Examining what happens in the shadows allows you to assess the exposure latitude (essentially the dynamic range) of the Raw files.

There are two factors at play here, which can't be entirely separated. Less exposure means less light is captured, meaning that shot noise is more visible. This means that the images would get noisier even if the camera was contributing no noise of its own. This theoretical 'perfect' sensor doesn't exist so, after a certian point, the noise contribution of the camera may also become visible.

Because shot noise is primarily dictated by the total amount of light captured, it is highly dependent on both exposure and sensor size. This means that, if sensor efficiency is similar, the results are only really comparable when looking at cameras with the same sensor size.

Compared with the Nikon D750, the Q is a tad bit behind at pushes up to 3EV. Beyond that point, the differences become more drastic, as the Q begins to show increased noise and banding relative to the D750. This performance should be seen in perspective, though: the Q offers slightly better performance than files from the 5DS R or 6D when it comes to pure noise levels at common output sizes. Note, however, that banding in the Q will ultimately limit the quality of these shadows, and is generally harder to remove with noise reduction or by downsampling your image to smaller sizes.

ISO Invariance

The exposure latitude test above includes both shot and read noise effects to give a real-world perspective on how far you're likely to be able to push the shadows in, for example, images of high contrast scenes exposed for the highlights. However, to get a clearer idea of sensor perfomance (and of the degree to which the sensor benefits from ISO amplification), we need to compare images shot at the same exposure, but different ISO settings.

Below we've used the same aperture and shutter speed at different ISO settings to see how much difference there is between shooting at a particular ISO setting (and using hardware amplification) or digitally correcting the brightness, later. Using the same exposure means that all the images were created from the same amount of total light (so have the same shot noise). This means that any differences in noise must be the result of read noise added by the camera.

As far as a 4EV push, we can see that the Leica is performing similarly to the Nikon, save for what appears to be hot pixels. To be fair, though, we don't know if this is currently an issue with lack of ACR support for hot pixel suppression for this camera. Beyond a 4EV push, banding increasingly overwhelms the image, as do general noise levels relative to the D750. This shows that the Leica Q isn't ISO invariant: it doesn't have so little downstream read noise that you can shoot at base ISO and push the results later in order to give yourself extra highlight headroom (cameras are generally capable of recording their widest dynamic range at base ISO). Instead it benefits from a little hardware amplification to lift all the data you might want to use above the camera's noise floor.

This is still pretty solid performance (its files appear to offer greater dynamic range than Canon's EOS 5DS R or 6D, for instance), but its shadow performance is fairly undermined by the hard-to-correct-for banding.