[Tig] 8k IMAX scans... 16k next

[TSassoon at aol.com]

In a message dated 7/24/08 8:53:34 AM, bfriesen at simple.dallas.tx.us writes:

> > http://www.studiodaily.com/main/news/headlines/9703.html
> 


Full of strange and crazy inaccuracies. For instance: "What’s more, to view 
the shots, they had to book time at London’s only IMAX theater." Would that be 
the BFI or the Science Museum theaters? Oh, wait. That's two. I've been to 
special screenings in both of them.

And Photoshop falling apart above 4K for matte paintings? Hello? I can think 
of reasons to not use Photoshop, but memory management isn't one of them. And 
the problem with higher-than-4K output for 65mm is that there are no laser 
recorders, so one is limited by the spot size on the CRT versus how many minutes 
you want to spend shooting a frame. DKP/IMAX solves this to some degree by 
shooting onto 5219 high speed camera stock, but I still see a lot of what looks 
like beam over-write to me (and they've agreed). And I haven't seen a 
higher-res output that was materially better than 4K. Usually it's worse.

The inefficiencies you speak of are usually regarded by the VFX house as 
competitive advantages. After a long time, they may decide to commercialize 
something, as in DD's case with Nuke. Generally though, the potential upside is much 
smaller than the potential downside for most shops, and it would take a lot 
of effort to make the proprietary tool into something releasable. I say this 
having personal experience in commercializing some of our plug-ins and shaders.

As for multiprocessing, third-party render management tools set up 
specifically for multi-processing are probably the way to go until the apps themselves 
are more competent. Gridiron Nucleo Pro for Adobe After Effects being an easy 
example. I imagine that their Shake helper was a RAM-disk cache manager to keep 
performance up. Shake itself should be fine in 32-bit.



Tim Sassoon
SFD vfx & creative post
Santa Monica, CA