P
Piotr Sobolewski
Hello,
I write a program which will stabilize a shaking video made with hand
camera. Of course it involves substracting frames, and it is substracting a
huge lot of one byte values.
The problem is that it takes a lot of time. To find out how was the camera
moved between two consecutive frames I have two substract that two frames
in all possible offsets (something like 40*40=1600 - I assume the camera
hasn't been moved more than 20 pixels in 1/25 s). And substracting two
frames is substracting 640*480*3 one byte values!
I guess it can be somehow optimised. For instance: using mencoder I can
encode videos to mpeg in real time, so probably they have some fast
method...
Does any of you have an idea how such substracting can be optimised?
I write a program which will stabilize a shaking video made with hand
camera. Of course it involves substracting frames, and it is substracting a
huge lot of one byte values.
The problem is that it takes a lot of time. To find out how was the camera
moved between two consecutive frames I have two substract that two frames
in all possible offsets (something like 40*40=1600 - I assume the camera
hasn't been moved more than 20 pixels in 1/25 s). And substracting two
frames is substracting 640*480*3 one byte values!
I guess it can be somehow optimised. For instance: using mencoder I can
encode videos to mpeg in real time, so probably they have some fast
method...
Does any of you have an idea how such substracting can be optimised?