Saturday, 29 December 2012

New Technologies of Digital Imaging

I recently wrote a post on CML asking the following:

"I saw The Hobbit in 3D at 24 fps at 2k, then walked into the next-door screening which was showing in 3D at 48fps at 2k.

So the first looked film-like and the second looked like old-style interlaced video - there was even a sensory and hallucinogenic lag in the image, mostly with regard to colour. People who buy 48fps argue that you should try to watch for 10 - 15 minutes to lock-in to the way you perceive the experience before condemning it out of hand.

Also as far as I understand it, instead of a 96th shutter, Lesnie shot the movie at a 64th shutter to add motion blur - and this didn't do anything to spoil the 24fps 2D filmic looking version when alternate frames were removed (being sharper than a 48th).

So my question is: Has anyone seen 48fps at 4k and if so, was the look filmic or video-like? 

I'm asking that question because it's my guess that 'film-immersion' works at certain 'sweet-spots' of the sensory experience and that because 24fps is one of those, then multiplying the factors could mean that either: 

a) a sweet-spot is disrupted if it's not a full multiplication of factors (so 3d at 48fps needs to be at 4k minimum to work) b) the next sweet-spot is a different multiple (96fps at 8k, or 120fps at 10k, or 192 fps at 16k - or something counter intuitive on an apparently different scale) c) after 24 fps the sweet-spot is way, way above those co-ordinates (on the basis that is a harmonic of the original)d) our senses film-immerse around 24 fps of image and 24 fps of black, regardless of resolution, and that's just how it is....

If there's anyone looking in from the production - you must have done shot some tests and screened variations - any comments? How did you prepare the 48fps and decimate the footage back to 24 fps?"

Someone wrote in response:

"I'm rather amazed and befuddled by all of these calculations and speculations as to the effect of framerates and sweet spots.  2D or 3D, the effect is apparent rather quickly, and this is nothing new. Oaklahoma! in Todd-AO (30fps) looks remarkably different to the CinemaScope (24fps) version which was shot along side it (a take with one camera, then a take with the other).  When the VariCam first came out I used it to demonstrate the difference between 24p and 30p, 48p and 60p.  The difference between 24p and 30p was easy to see for all.  You can think of it whatever you wish, but the moment you get to 30p the effect is very "video-esque" in motion, at least to a brain used to 60i American television.  There's no magical formula of shutter angle and 3D immersion which will change this.  Refresh rate is refresh rate and the mental connotation is, well, whatever the viewer brings to the table.  One can give it a try and decide if it is interesting or acceptable, but the effect is the effect and it's really that simple.  There's no "training the audience," no "finding the magic combination," no "filmmakers not using their tools properly."  Either it is liked by the audience or it is not.

And to a great degree, I think this is true of 3D in general".

I felt I had to reply:

"Mitch and Mike: we're ok with numbers aren't we? After all that's an aspect of much of what we do.
I didn't know the Hobbit was finished in 2k, so that was worth writing the post in the first place for, as I found something out - it certainly kills the issue of seeing a '4k' version at 48fps.

'There's no "training the audience," no "finding the magic combination," no "filmmakers not using their tools properly." Either it is liked by the audience or it is not.....And to a great degree, I think this is true of 3D in general'.

I get it that there should be a resistance to pr style thinking and I wasn't really interested in the 3D issue, as my interest is with audience immersion (we use light and camera movements to underscore dramatic narrative and deepen audience engagement - why not use new capacities in the medium we work in?)
I'm very privileged to be working on higher dynamic range capture and display and when you see this actually working before your eyes there's a sense of seeing three dimensions, which comes through without the tricks of standard stereopsis. The response after seeing this new imaging technology for the first time is: 'It's like looking through a window'. When you see an HDR image in HDR display space, the sense of it being plastic and unreal goes (mainly because up till now HDR images were seen in non-HDR display space and the audience didn't like what they saw). The possibility for really amazing lighting is there, because the display space is approximate to the eye-brain pathway.

True, up till recently each image has been captured using 7 or more exposures in each frame and so has been still-image or stop-motion - but we're now perfecting moving image HDR streams instead of stepped still images. At the capture stage data bottlenecks are becoming an issue as we're generating huge amounts of data (could be up to 1 terabyte per minute next year as we'll move into high fps hdr - so there'll need to be developments in every part of the chain - that'll explain my OCD interest in numbers then). Most importantly though, we'll be looking at what kind of content works in this new area.

Call me old fashioned or even OCD, but I am interested and curious about what can be done outside of what's currently liked by an audience, and I'd prefer this information to come out through CML, within the community that creates moving image art for a living - not only that but involve members in trying to do this. 'Course, we may not manage to pull it off, but I'm quite happy to go down in flames trying."

So reader: What do you think?