Scientific imaging cameras typically produce images with 16-bit (0-65535), 14-bit (0-16383), or 12-bit (0-4095) data range. Subsequent processing such as stacking can increase the bit depth. As an example, if your camera produces a pixel value 16383 that number can be stored in 14 bits. But if you add together 4 images with the same value, you get 65,532 - a number that requires 16 bits to store. Similarly if you average numbers, you will get fractional ADU values, which need to be stored in a floating point number.
By default, images captured with a camera are saved in FITS 16-bit format. If you perform processing such as stacking, you should store in FITS 32-bit IEEE Float format. The Stacker can be set to mark images to be saved in either 16-bit or float formats; it is strongly recommended to always use floating point to avoid losing data.
You may wish to save an image into an 8-bit format such as JPEG or PNG, for posting on the web or viewing in other software. Since those formats have lower bit depth, you need to adjust the data range before saving; in fact failing to do so often results in completely blank images, as the lowest pixel in the image could easily be higher than the maximum 8-bit value of 255.
The File menu Save As command warns you in a text box when the data in the image exceeds the capabilities of the file. If this happens, make sure the Screen Stretch is set appropriately, and turn on the Auto Stretch check box. The image will automatically be scaled so the saved image will look just like the image on-screen. Note that when this is done, data precision is permanently lost. You should always save your images in FITS format first before saving into an 8-bit format.