The Hobbit Filming at 48fps

An anonymous reader writes "Peter Jackson has announced via his Facebook page that The Hobbit is being shot at 48 frames per second, ameliorating the '3D headaches' that many viewers have complained of in the last few boom-years for the format. Film has been shot and projected at 24fps since the 1920s, with the exception of Douglas Trumbull's 60fps 'ShowScan' format, used for the Universal Back To The Future ride, amongst others. Jackson himself predicts that the widespread adoption of 48fps workflow could not only improve the 3D but also the general cinematic experience, though it may earn itself some backward-looking critics. But until digital principal photography completely usurps celluloid, this may be good news for Kodak, who now have even more reason to lament the death of Stanley Kubrick."

Comments

imax & imax dome

[bradgoodman]

On particularly large screens - the relatively "slow" frame-rates used today are quite troublesome. For example, say your shooting video out of a front of a plane on an imax dome screen. When the plane banks - even if it does relatively slowly - since the screen is so large, you see a lot of "jumpiness" - as there may be several *feet* in real-world on-screen distance between an object's position in one frame vs. another. I've been complaining about this for years. It would be nice to see higher frame rates in formats like this.

Re:Wrong problem anyone?

[Rei]

You know, I wonder when we'll ultimately just drop the concept of "frames" and switch to temporal-tagged packets of image changes, without requiring a full image to have been acquired simultaneously. Aka, your CCD doesn't accumulate photon counts, but photon rates. The readout from the CCD returns the delta between the current rate of activation and the previous activation rate. For a CCD polled thousands of times per second, for most pixels, that would be near zero, and that pixel is declared "unchanged" and ignored. The pixels which have a statistically significant changes are returned to the camera as ID/rate pairs, and are all bundled together with a time tag, processed, and compressed. Then it's a trivial matter to assemble them into whatever frame rate you want, it makes it much easier to do high quality slow motion, etc. Our insistence on accumulating all data into (proportionally slow) "frames" during the recording process is throwing away data.

Of course, this would require some significant hardware and video format changes, plus different approaches to compression, as the data you're reading is loosely packed instead of densely packed. Good compression approaches would take into account the strong regional correlations between pixels reporting changes in light intensity.