January 1st, 2023, 14:33
January 1st, 2023, 18:07
January 2nd, 2023, 18:00
January 3rd, 2023, 6:11
January 3rd, 2023, 9:51
pepe wrote:i was thinking about this a few weeks ago with my son. If you have a single underexposed image, a significant amount of information is missing, the image has a lot less brightness levels, which will result in the same amount of discrete levels after amplifying it. One can try to add that info manually but i would not call this image processing, it resembles more to some kind of an art. It looks good at the end and might be close to the original subject, but it is still an artificial image.
I agree that it is a lot different if you have multiple images, from which it is possible to get more info and restore a better single image.
January 3rd, 2023, 17:46
January 3rd, 2023, 18:10
January 3rd, 2023, 19:58
January 3rd, 2023, 20:50
pepe wrote:the 'just isn't there' was not about what you or anybody can reveal by adjusting the levels. I meant that there is no detail in between those levels. I never meant to say there is no data at all if we can't see it because of being underexposed, just that a lot of the details are lost because the very low number of levels actually showing up in the image.
btw, the before-after pic is also a cheat because the image downloaded from your first link already shows something while his 'before' pic shows full black except the window. So we are definitely talking about different things, i took his 'before' pic as a basis of my comments.
I, as anybody else, was able to make the pic you linked a lot more enjoyable in 5 minutes of adjusting but i'm afraid his 'before' pic was darkened intentionally
this is what you get if you start adjusting the levels of his 'before', with its histogram. Fairly obvious that pic contains very few levels, a lot fewer than what you linked as originals...
So where is the big deal? presenting a manually degraded image as original while using another one as a basis of our transformation????
January 3rd, 2023, 21:45
January 3rd, 2023, 21:58
What's you're using is an already processed image, probably some lower res JPEG for web use = missing the point spectacularly. How would the page load when he'd use the almost 1.3 GB original scan?
January 3rd, 2023, 22:16
it is not about the resolution, what he showed on the left is not the original image and not in the sense of resolution (who cares about resolution?) but in darkening it significantly, and marking the darkened image as original. This is where i feel cheated.
this is what i made quickly out of the original you linked:
but the previous one i made from his 'before' pic which was definitely darkened significantly, not the original in the sense of lighting, not talking about resolution which is not really important right
January 4th, 2023, 10:05
low res, low quality.
January 4th, 2023, 10:45
pepe wrote:i am noob as well when it comes to photo editing, just used mask. I am pretty sure it can be done a lot better by applying some lighting and colouring, but i am not a master of this thing, only have the ideas but not the experience how to do it right.low res, low quality.
true but currently our problem is not the x-y resolution but the very low dynamic range of colors. And his 'before' image has a lot less levels present than the original we used, there's where i suspect cheating...
btw, for the above pic i used the small (800kb) version. Of course, if you have a large TIFF with lossless compression, there will be far less noise in the result as well. But for demonstration the smallest one is perfectly ok.
pepe
January 4th, 2023, 11:38
January 4th, 2023, 17:12
Powered by phpBB © phpBB Group.