Strange behavior (for me) of the Exposure module.
Two identical photos (exposures).
They look the same.
What differs is the light metering and exposure compensation.
One photo has full-frame metering (Sony/Multi) and a +1/3EV correction. Darktable knows that metering mode = multisegment.
The second photo has metering set to highlight sparing (Sony/Highlights), and then we have a +2.7EV correction.
Darktable shows that metering mode = other
The same exposure on the same scene/lighting on the same camera/lens should give the same result.
Would you mind confirming the exposure conditions? You should be able to find extracted exif information on the left panel in DT, or on the camera itself if the images are still there.
I donât know Sony cameras, but âpreserve highlightsâ should tend to lower overall exposure.
Iâm wondering if dt is reading exposure compensation correctly.
There are some settings where the camera effectively lies about the ISO used. Different brands have different names for these things, with Sony calling it Dynamic Range Optimizer. Maybe preserve highlights also does it. With 5.4 DT will be able to compensate for many of these cases.
Darktable applies the compensation for the value set on the camera Ev compensation dial. Thus we are looking at one picture here that is over two stops less than the other.
Unselect the exposure compensation. I sometimes wonder about that default, because if I applied Ev compensation, I knew what I was doing (or trying to).
metering mode? Interested to know if Iâm wrong, but I donât think that dartable takes any notice of that at all. (???)
How I took both photos: Aperture priority mode, f9, ISO100
1.
I set the light metering to Multi.
I set the exposure compensation to show a zebra pattern in some places = +1/3EV
I noted the exposure time = 1/250s
I set the light metering to highlight (respect). The camera immediately darkened the image to ensure the highlights were correct.
I set the exposure compensation so that the exposure time was the same as before = +2.7EV
So we have two photos that look the same, with the same exposure parameters, except for the exposure compensation.
Darktable correctly reads the exposure compensation (+1/3 and +2.7EV), and finally, the Exposure module shows the second photo (+2.7EV) as significantly darker.
When we turn off âcompensate camera exposure,â both photos look the same.
And thatâs why you see a difference in darktable: darktable doesnât âseeâ that the camera darkened the image, only that you applied +2.7EV compensation. In short, the camera was lying and darktable has no way of knowing that.
Which is also as it should be: you applied a correction to get the same visual result, and you tell darktable not to undo that correction.
But darktable canât know what you were trying to do.
If you want to actually modify the image by using EV correction, darktable shouldnât touch that correction.
If you applied a correction for âETTRâ (either + to get better data in the dark areas, or - to protect bright areas), you do want dt to correct that⌠The default setting assumes ETTR.
This is literally what the check box is designed to do, isnât it? I think there is a misunderstanding here.
The manual states in the section on the exposure module, under âcompensate camera exposure (manual mode)â:
Automatically remove the camera exposure bias (taken from the imageâs Exif data).
So if you took a photo with +2.7EV exposure bias, checking this box will remove 2.7 EV. Metering mode is not part of this process at all.
Iâd imagine the point is if you purposefully (manually) underexposed to protect highlights, or similar situations. Itâs not an option you always want to use, and what youâre demonstrating here is that itâs functioning as intended.
