I read about a method over on the FredMiranda.com forum for "Auto ETTR" for the Sony A7RIII using the new Highlight metering mode that is also available on the a6500, so I decided to run a quick test to try it out. ETTR (expose to the right) is a method used to obtain the best possible image quality from a single exposure by intentionally over-exposing the image as much as possible without clipping the highlights. In some situations, this method would actually under-expose an image if needed to prevent the highlights from being lost. In theory, it sounds simple, but since our cameras don't show a true histogram of the RAW image data, it can often appear that an image has been overexposed when in fact the highlights could be pushed further in the RAW image. Thus, any new feature of a camera that can simplify or improve the process of ETTR could be very useful.
For Auto ETTR, Fred's method involves setting the camera's metering mode to Highlight, then adding +2 stops of Exposure Compensation. He said in his testing with the A7RIII that +2 was very safe and would not over-expose the highlights (http://www.fredmiranda.com/forum/topic/1519071/9). My question is, how does this method work with the a6500 and is +2 still a safe value? In the past with the a6000, my method has typically been to use about +0.7 exposure compensation with Multi metering to push the exposure just a little.
This morning we had some nice clouds to the east, so I stepped out on the driveway to snap a few frames with various exposure settings. I used the Sony a6500 with Zeiss 24mm f/1.8 lens shooting in Aperture priority mode, ISO 100, and f/5.6. I took a series of images starting with a base exposure using default Multi metering with no adjustment, then switched to Highlight metering with exposure compensation of 0, +1, +2, and +3 successively. This test is very unscientific, and in fact, the light changed considerably while I was shooting. But let's look at the results anyway. All images processed in Lightroom 7.3.1 using the Adobe Color profile. I set Highlights to -100, Shadows +70, White to Auto, and adjusted Exposure and Blacks manually to try to match the base image. (And after publishing this article, I have realized that this was a very poor scene for a test of ETTR. But it is a good test of highlight recovery.)
Looking at the series of images, it is difficult to compare the Multi metering exposure to the others because the sun was behind the clouds, so the backlight was not nearly as intense. Even so, the sky was extremely bright yet the image retains good detail in the clouds. However, I'm not sure what the camera would have done if the sun had been directly in the scene. After I switched to Highlight metering, the sun started to break through creating a much stronger backlight. Looking at the four Highlight metering images, only the +3 exposure appears to be blown out where the clouds around the sun are pure white. I would say the +2 exposure is just about right for the highlights. So Fred's method appears to work for the a6500 as well as the A7RIII.
However, looking at the shadows is a different story. I'm seeing a lot of noise, even in the Highlight +3 image, but I also recognize that this was an extremely high contrast scene and probably exceeds the capability of the sensor to capture the full dynamic range. Further testing is definitely needed, but here are the results zoomed in at 200% to show the noise.
Noise in the base (Multi) exposure is OK, but is really excessive in the Highlight metering exposures, even at +2 and +3. But again, I think this scene was pushing the sensor too far. I'll do further testing, but for a backlit scene like this one, I would typically be bracketing for HDR.