I'm watching TV and I just heard something along the lines of "The files have been wiped from the server and there was no sign of a DDOS attack. Whoever erased those files had a backdoor.".

With them being at this profound level of being dumb...

Someone might just have entered through the backdoor.....and deleted their shit

I work with cameras and images.

"we got this lowres video from the traffic cameras. I'm using a special algorithm to increase the resolution to 4k!"

THAT IS NOT POSSIBLE!!!

@CWins oh wait really!! What about x50 zoom and enhance surely that's possible. (sarcasm incase you missed it)

@CWins "Can you enhance that?"
Biep biep biep - Done.

"Ok, can you run face recognition on him?"
Biep biep biiiieeep - His name is $genericRussianName!

"Our writer was too lazy to come up with an interesting way for us to catch the criminal so can you just please scan all the security cameras in the whole city for his face and give me his position in 3 seconds?"

@hash
That, plus the reflection on the car on the other side can be combined into a 3D model.

@CWins Are you sure?

https://petapixel.com/2016/11/...

@psfr937 That might work for distinct objects, but it won't be able to make blurry number plates readable or enhance a face consisting of 20 greyscale pixels.
The problem with either of these is that, unlike eyelashes, they are unique for every person and can therefore not be estimated. The same way you could not estimate somebody's fingerprint.

But still, the project looks really interesting!

@CWins why didn't my copy of Photoshop come with an "enhance" tool?

@hash that would have been a good place to use the <sarcasm> tags.

@psfr937
An article worth reading. Thanks for the link!

But yes, i am sure. I did my diploma thesis on a superresolution topic and i've been working on machine vision since then. There are some really cool things can that can done with superresolution and "educated guesses" and if i remember correctly, the series "Numb3rs" did show superresolution in a way that uses multiple frames of a video to create a better image. This works, because each frame of a moving object contains slightly different data while other data is lost in sampling. There is a limit this approach and details that are simply to small for the resolution cannot be brought back because they are not contained in the data.
(1/2)

(2/2)
Googles "educated guesses" is great for increasing image quality, but it utilizes data that should be there accoding to learned data. That's fine for something like a metal border or those barcodes that suffer from aliasing, but if a photo is too low res to show the eyes, there's no way an algorith can find out the eyecolor of that specific person.

@Jop-
I would give my credit card number for that information :-D

So can he hack facebook?