I work for a Planetarium. We have a few domed theaters. We are in the process of revamping one of them to be high-def, and hope to overhaul the authoring process at the same time.
Recently, we received a donation of a number of projectors capable of displaying 720p, 1280 x 720 res.
We are working on a solution for synchronizing playback so they're frame accurate with one another, and edge-blended. They would produce a seamless band of HD video around the entire perimeter of the dome, extending up about 22 degrees from the 'horizon'. The output resolution of the virtual 'canvas' we would be authoring on is eight projectors 'wide', and circular (eg left edge of projector 1 needs to mate up with right edge of projector 8), and it spans 10,240 pixels.
Despite this somewhat intimidating resolution (hmm, that's four apple 30" cinema displays side by side... wonder if I can convince the boss...) we would still very much like to be able to produce media in-house for playback on this system. Preferably with simple keyframed animation and layers, across one big 'canvas'.
The source media would primarily be large scans and digital photographs, some possibly in RAW format, plus some pre-rendered animations and the occassional video clip. Right now we bring in DV from a 3-chip 1/2" CCD ENG camera and some Beta SP stuff but it's all SD; we'd like to get into HD res video in the near future, though.
After laying it all out, compositing, animating it, crossfading, whatever... we'd then want to 'slice' this huge virtual canvas into eight 16:9 ratio, 1280 x 720 segments, encode those into MPEG2 transport streams, and load them onto the HD playback devices. Is Shake well-equipped to handle this 'slice'? It seems like I'd have one FileOut node for each virtual 'camera', each one having its own piece of the canvas to display. (Am I on the right track here? Still learning the metaphors and terminology.)
I've been looking at Shake as a possible 'solution' for this workflow, primarily because of its unlimited canvas size (Final Cut is limited to 4k wide), distributed rendering capabilities (we've got a few 'clusters' worth of machines to render on), its 'proxy' abilities, and so forth. But I have some reservations.
Shake looks to me like it's meant for compositing individual shots, whereas I want to tell a whole story. (30+ mins.) But I guess if that's the case, then we could get around this by doing each 'scene' as its own script, 'slicing' them all, and then cutting them all together in Final Cut at the 720p res (frame accurately? Sure, I guess...) before encoding?
I'm also concerned about the 'wrapping' that we need to simulate between 'tile 1' and 'tile 8', but we've done that before in our earlier, more archaic production method (think cross-fading a LOT of slide projectors...), so we can probably handle that.
Can anyone offer suggestions? e.g. is Shake the right tool? If not, what would be better for this workflow? If it is, are there things we'd really want to pay close attention to to avoid bottlenecks/pitfalls? Are there specific third-party plugins that we'd want to get?
Thanks in advance...
Thom Brooks
Adler Planetarium and Astronomy Museum
Chicago, IL USA