The on-board camera will be taking photos in quite unusual conditions, far different from the ones it's been designe for, and therefore unusual problems will arise.
Of course, no-one will be up there pressing the shutter to take photos. You have here two approaches: the on-board computer tells the camera when to take a photo, or the camera decides on its own what to do. Being an avid user of CHDK, I prefer the second approach that will make the camera more independent from the computer. For example, the camera will be able to switch from taking photos to taking videos, might decide to shoot less photos during the ascent phase and more at high altitude, and so on.
The camera hangs below a balloon that is carried around by winds often in excess of 150 km/h. This is not the ideal situation for shooting and measures have to be taken to prevend badly blurred images
At high altitude, we have a strongly contrasted scene, with the earth below us under a strng sun (especially if there are a few clouds), and the sky above being completely dark. Also, the camera oscillates quite strongly, alternatively pointing downwards and upwards. If often happens that during the slight delay between the camera metering and the actual photo being taken, the camera moves from a bright to a dark area, and vice-versa. The risk of under- (or over-) exposed photos is substancial, and usually a significant number of shots are unusable because of this problem
You can't send up there a really good camera, and this for a couple of very good reasons. First of all, the risk of losing the capsule is significant, and this would mean losing an expensive camera. Second, with the complete payload usually in the order of a single kilogram, you are forced to use small point-and-shoot cameras that often have shortcomings and limitations