First, you have to enable the 'ettr' module. Once enabled, select the Expo->AutoETTR setting. For timelapse chasing sunsets/rises, I set it to Always ON so that it will determine what to do on each frame. For sunsets, I also set the 'Slowest shutter' setting to what works in that environment (city scapes ~2secs, dark sky locations ~20-30secs). The concept is trying to push your exposure to the limits of the digital sensor. This enables the shadows to receive extra exposure while counting on the fact that you can recover a certain amount of highlight that would normally look over exposed. In fact, if you shoot JPEG images, it will be over exposed. Only shoot RAW. ML evaluates the exposure of the current frame, and then looks to see if it needs to adjust it up/down. As the scene gets brighter, it will speed up the shutter speed. As the scene gets darker, it slows down the shutter speed. If the 'Slowest shutter' speed is reached, ML will then try bumping up the ISO to keep the exposure the same. My mkII stops at 1600.
ETTR is an acronym for 'expose to the right'. It's fine for sunsets. But are portraits trickier because the sensor's gamma curve is optmized for skintones falling within a certain IRE range? I come from the world of Arri and Red where this is a thing - but perhaps the high dynamic range of stills makes this a non-issue.
I've had this conversation with portrait guys that think I'm crazy for pushing the exposure that hard. Granted, I don't shoot a lot of timelapse with models ;-) I do know that pulling skin tones back out of shots that were 'accidentally' over exposed was tough/near impossible. As with all things, each camera has its unique qualities/abilities. Whenever I talk ETTR with people, I always suggest testing the limits of each camera body and the ability in post to recover the highlights. Test, test, test, and then know those limits so you can get the max ability from each piece of gear.
The crazy thing is that I come from a video engineering background, and image acquisition with just a histogram is still foreign to me. Give me a good waveform for exposure and vector scope for color, and I'm much more confident about the image. I want my camera cart to look like a DIT station!
Yeah, I mean I think it would be such a minimal difference in recovered highlights with a raw image where you've ETTR. I think testing the sensor limits is a great philosophy, but also letting go and focusing on the 'artistic' subtleties over the technical subtleties of an image becomes more important.
Would be interesting to a/b your ETTR and 'correctly' exposed images to your friends without letting them know.
Sensor response is pretty much linear. That's why you want to shoot in RAW which gets you file without a tone function applied. Whereas with JPEG, it's not only applying a tone function but a contrast function as well which aggressively tosses highlight and shadow detail.
