deblurring software presented @ Siggraph. Next we'll see this technology built into digital cameras...
If anybody is not up to download 54Mb, here is a plain page to start. Executive summary:
A pair of images, one with correct exposure, but blurred due to the camera shake, another way underexposed, hence sharp but noisy, is used to produce a surprisingly nice final result.
Natural caveat: requires two different exposures. However, if paired with the proper hardware (in-camera, like Doug suggested.) it could become a life saver soon enough.
That would probably be a little too much computing power to just add to a consumer camera. The higher end models wouldn't need it because the people who buy and use those don't need it.
And you need two pictures of the same stationary subject. Not always possible.
"A photo is like a hamburger. You can get one from McDonalds for $1, one from Chili's for $5, or one from Ruth's Chris for $15. You usually get what you pay for, but don't expect a Ruth's Chris burger at a McDonalds price, if you want that, go cook it yourself." - me
That would probably be a little too much computing power to just add to a consumer camera. The higher end models wouldn't need it because the people who buy and use those don't need it.
And you need two pictures of the same stationary subject. Not always possible.
Not exactly.
Today's cameras Digital Signal Processors have enough horsepower to do some extra work (c.g. MarkIII with two CPUs).
And as to the two exposures: it's feasible to devise a mode when data would be read off the censor and stored away shortly after the exposure has been initiated, thus providing a "underexposed sharp" frame. Then exposure would continue normally for the "exposed blurry" frame and finally both will be pocessed for the final result.
I'd love to have 50D with this mode:-)
Comments
If anybody is not up to download 54Mb, here is a plain page to start.
Executive summary:
A pair of images, one with correct exposure, but blurred due to the camera shake, another way underexposed, hence sharp but noisy, is used to produce a surprisingly nice final result.
Natural caveat: requires two different exposures. However, if paired with the proper hardware (in-camera, like Doug suggested.) it could become a life saver soon enough.
And you need two pictures of the same stationary subject. Not always possible.
Today's cameras Digital Signal Processors have enough horsepower to do some extra work (c.g. MarkIII with two CPUs).
And as to the two exposures: it's feasible to devise a mode when data would be read off the censor and stored away shortly after the exposure has been initiated, thus providing a "underexposed sharp" frame. Then exposure would continue normally for the "exposed blurry" frame and finally both will be pocessed for the final result.
I'd love to have 50D with this mode:-)