(This article is a supporting part of our ongoing testing of low-light camera metering reliability)
The Sony A7 Mk. III is our current champion for low-light metering abilities. It first fails to compensate at least 2/3 stop for a 1 stop decrease in light at EV-6.4, which is approximately equivalent to the light level provided by an 18% moon. At scenes darker than EV-8.1, its images are over one stop darker than those auto-metered at EV0. Compared to other cameras, the metering of the A7 Mk. III abruptly “gives up” beyond about EV-6. The manufacturer reports that its metering limit is “-3 EV at F2,” which is significantly brighter than what we measure.
The Sony A7 Mk. III stops down the lens to the user-set aperture at all times, not just when taking a photo. In other words, if you are using an f/1.4 lens but have the camera set to f/4 for improved sharpness, the lens stays at f/4. The metering threshold is three stops worse than if the camera was set at f/1.4! Sony keeps the lens stopped down for two reasons. First, it gives the user a real-time preview of the depth of field. Second, some lenses have significant spherical aberration, and this causes the plane of focus to shift when stopping down. If you had such a lens and focused with it wide open, the resulting image would be out of focus when you stopped down and took the photo. However, cameras from most other manufacturers give the user a “depth of field preview” button, and this stops down the aperture when pressed. The focus shift could be solved by not selling lenses with enormous spherical aberration, or focusing stopped down and metering with the lens wide open, or having a lens-dependent lookup table for focus shifts. The problem is solvable, and Sony’s solution is just lazy engineering. Not only is metering worse because of keeping the aperture stopped down, but autofocus is slower too. The Sony A7 Mk. III has an option called “live view shooting: setting effect off,” but this doesn’t solve the stop-down issue.
The Sony A7 Mk. III has a variety of options for when to increase the ISO vs shutter speed in Auto ISO mode. This is very helpful, since you can require a (for example) 30-second exposure time before the ISO increases over 100. Alternatively, the ISO can be set to keep the shutter speed at least 1/focal length in order to minimize camera shake.
The Sony A7 Mk. III has the option of an all-electronic shutter, making it a good choice for nighttime webcams which take a million photos per year. In auto-ISO mode, timelapse has very little flicker caused by settings changes; I suspect that it steps its settings in increments smaller than 1/3 stop, though this is not reported in the EXIF data.
The plot shown above was created using the usual EF-mount Rokinon 50mm f/1.4 adapted to Sony and set to f/2.8. The aperture-dependent metering phenomenon was tested with a Sony 24mm f/1.4.
Integrating Sphere & Camera Metering Test Project