I am working to determine the orbit of a satellite based on a long exposure photograph, and need precise observations.
The EXIF standard describes the
DateTimeOriginal tag simply as "the date and time when the original image data was generated." It gives no guidance about what event (e.g., shutter released, shutter open, shutter closed, sensor read, post processing) should be used to determine the value.
As mentioned in a comment, the best way to find out how your camera defines it is through experimentation. Synchronize its clock to one you can observe, record the start and end times of a long exposure and see which one ends up in the EXIF. Note that the result you get will only be valid for that one camera with whatever firmware it was running.
Experimentally (on my EOS 70D), this is the beginning of the exposure, and not the end.
- this seems truncated to the second
- it depends how accurate is the time of the camera (before doing this I carefully set the time on my camera, but I doubt I can do better than half a second)
- ... not speaking of clock drift if it hasn't been set recently
IMHO a better method is to trigger the picture from an external device than can have a really accurate clock (Raspberry with NTP clock, for instance)
Just did a bulb exposure with a Nikon D850 of over a minute and a half (to make sure that time, during the exposure, chaged the time not just in seconds but also at least in a minute).
In this camera, at least for bulb, the recorded date time is the initial moment the shutter pressed, when exposure start, not when it’s released and the exposure ends. For other models and makers it may vary.