You are not logged in.
Hi!
Some year(s) ago, I've seen several articles + demonstrations of different techniques to find suitable potential still images in videos and combine similar frames into images with increased resolution, reduced motion blur, etc. Discussed among other things were camera motion tracking, recognition of the same scene with different lighting, different algorithms to combine frames that show essentially the same scene but with different noise over time into a bigger image or one with reduced noise... also combination of "low light noise frames" into "simulated high exposure time photo". There were several screen-casts of a (rudimentary) tech demo that was running in a QT decorated GUI on an ubuntu desktop, and there was some sort of API to download and play with (but I didn't). I think it was on some university website, but I don't remember many details.
It was all very experimental and messy and I wanted to take another look at it later, to see if anything useful came of it. Now I can't find anything about those projects or techniques any more.
Does anyone know anything about applications or even just API's that do something like that?
Last edited by whoops (2015-08-05 17:55:58)
Offline
Not sure this is what your looking for but VLC does have Video effects and filters in the tools menu..
You could look at Natron
Offline
Thanks, but not what I'm looking for. Part of the problem is, I don't remember what it all was called.
Let me try again:
I'm looking for an application (or API) that... can (help) locate "scenes" based on how similar frames are, compensate for small camera movements + lighting changes and crop + rotate frames that qualify for frame stacking to create a superresolution image of that scene?
There was an article there that explained how some different "pixel jitter" has to be introduced into each frame before they are combined and slighly blurred. And stuff...
Offline
As for super-resolution i can't say i know of any solutions on linux... but i would still look at Natron
Offline