I really need help fixing these photos. My mom (71 yrs old) had seeing the aurora on her bucket list and so we went to Norway (from DALLAS!) to see them. We did, luckily, but the photographs all turned out black! If I play with the histogram, the aurora is there, but it is terribly noisy. I'm a bad photographer and not great at photoshop, so I have no idea how to fix this. Please help!Yep, these were point and shoot cameras.  Leave it to amateurs to mess up a once in a lifetime opportunity.

  • 8
    You can't fix what isn't there. If this was shot in some RAW format(not JPEG) there might be some hope. You might be able to take some abstract "artistic" take on the image but it will never look as it did in person.
    – dpollitt
    Jul 11 '14 at 16:41
  • 2
    Watami, You mention "photographs", plural. If you took a few shots of the same thing from the same spot, I might be able to help. I would simply "stack" the images. You can easily do the same in Photoshop or other software, too. The good news is the aurora is pretty amorphous, so you don't have to perfectly align edges to get reasonable results. You just have to be close. Also, make sure you always save an un-edited original set of the images before you tinker with them.
    – B Shaw
    Jul 12 '14 at 3:04

If you open the image with any editor, you can see in the histogram (below) that it only uses (in a significant proportion) 12 of the 255 available brightness levels. This means that it only has 12 shades to represent the image that it captured.

enter image description here

Switching to the logarithmic histogram,

enter image description here

you can see that actually there is more, but the most significant part, representing more than 99.9% of the image, is between levels 0 and 16 of each channel and 99.8 of the absolute greyscale.

(The red part on the right of the histogram is just the date/timestamp)

You can still do something to "push" the exposure and get an idea of the picture:

enter image description here

But unfortunately it can't get much better, since the information is simply not there.

  • I would recommend a "push" of the date time stamp :-P
    – dpollitt
    Jul 11 '14 at 18:38
  • @dpollitt whoops, that's the reason for the red pixels. Editing...
    – clabacchio
    Jul 11 '14 at 19:32
  • 4
    I just get a kick out of when people have timestamps on their images... I thought digital photography made that obsolete but people still seem to love it.
    – dpollitt
    Jul 11 '14 at 19:33
  • @dpollitt consider that many people don't know how to read metadata/exif
    – clabacchio
    Jul 11 '14 at 19:34
  • 2
    If you click on an image in Windows, Explorer will tell you the capture date/time. It's pretty easy!
    – dpollitt
    Jul 11 '14 at 20:19

The extra information simply isn't in this JPG file since the pictures is so grossly underexposed. Here is what you get by expanding the dark areas (ignoring the silly date/time stamp and the bright spot at lower left) to the full black to white range:

This looks so grainy because only a few intensity levels are expanded to cover the whole range. The only way to deal with this now is to go back to the raw file and do the same transformation I did. With 12 bit raw data, there are 16 times more levels, and with 14 bit raw data 64 times more levels available. Still, it looks like your camera has significant pixel noise, so even that will look "grainy".

The right answer is of course to not get yourself into this mess in the first place. Didn't you take even a quick look at the pictures in the camera on the spot? Also, it should have been obvious the scene was very dark, and you needed to set up your camera carefully to make sure it captured what you intended.

Another thing to do is to always post-process from the raw file. At least you can still do this, unless of course you deleted it, in which case your images are gone for good.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.