dual iso should really be combined with an ettr/ettl option to expose the first image to one side and then automatically figure out the required ev spacing for the 2nd shot, in essence minimizing the required dual_iso ev spacing.
Almost working:
https://bitbucket.org/hudson/magic-lantern/commits/eb3cc38d1870How it works:
- remember that ETTR has SNR limits for midtones and shadows (it will try really hard to meet or exceed that SNR).
- if dual ISO is not enabled, it may have to clip many highlights in order to meet that SNR. If you turn off SNR limits, it will underexpose, but keep all the highlights.
- if dual ISO is enabled, ETTR will adjust it to meet or exceed the SNR you have set, and keep all the highlights (except the specular ones, handled by "Highlight ignore")
Problem: ETTR is iterative, and it relies on the previous image (where it measures the white point, some SNR numbers and stuff like that). SNR seems to change quite a bit with dual ISO (even if it shouldn't); need to investigate why. So, at this point, ETTR with dual ISO may be a little undecided: it may say 100/800 or 100/1600.
How to use it:
- Make sure you have raw zebras and histogram enabled (so you know what's going on, what's clipped, what may be aliased...)
- Enable Auto ETTR (Always ON) and dual ISO.
- Point the camera at your scene and take a few test pics.
- If the exposure doesn't converge, try underexposing a bit. When the image is overexposed, it has to guess by how much (and it doesn't always work). When it's a little underexposed, it can compute exactly what settings are needed. When it's heavily underexposed, noise will affect the metering.
ETTR will try to:
- use a low ISOs and a low EV spacing
- meet or exceed the selected SNR minus 0.5 EV
- do not clip more highlights than you have selected for "Highlight ignore"
- do not exceed max auto ISO from Canon menu
So, if that SNR can be met without clipping highlights, the scene does not require dual ISO. If not... it does.
Enable debug info to show the math behind it.