Hard to comprehend that much footage.......amazing
Hard to comprehend that much footage.......amazing
Jim, did you ask Peter what he thought about the opinions, from critics of early screenings of the movie (Mainly the 48fps negative feedback)?
The camera's have already proven to be a work of genius. You mentioned he's happy. That would be because he knows how it's all gonna turn out.
1,000 Terabytes is a petabyte: http://arstechnica.com/information-t...-tape-storage/ .
1,000 of those is an Exabyte: http://en.wikipedia.org/wiki/Exabyte.
I remember the first 1 gigabyte hard drive's from around 1993 - 1994. I wonder where we'll be in another 20 years.
I'm reminded of the time in 2003 when Kodak sent director Dylan Sanders a case of (fairly expensive) Cristal champagne when they passed 2 million feet of film on Charlie's Angels 2...
We still got exabytes, zetabytes, and yottabytes above petabytes:
We'll need drives this big when we're dealing with 48K 120fps 4D footage in ten years.
With performance capture to some degree we could theoretically save storage. Texture once, render multiple frames. Only need the "meta data" e.g. Animation for every subsequent frame which is relatively small to say a 48k frame.
Data rates have been climbing steadily for years, going back from analog 525 composite to component analog to component digital to HD to 2K to 4K and on and on and on. I don't see this changing. People's expectations are raised. Imax theaters are making a lot of money, so it's clear the public will pay more money to see bright, sharp, beautiful pictures, in 3D or not. And that takes a lot of data (whether measured in terms of film footage or data).
I think the days are soon coming where even a modest post facility in a major metro area will routinely have to have several petabytes of fast network storage. I don't see this changing. The real question to me is, "how many K are enough?" I don't have an answer to that... but then, I was perfectly happy with how Avengers looked, and I'm sure Disney was happy with the box office.
In the Hasselblad to Red comparison we saw a stills photographer get his eyes opened to being able to capture just the perfect moment every time - he just had to walk through the footage to pick it. He found it a bit strange to get his head around the paradigm shift and the way it changed his workflow.
What PJ does is the same thing applied to completely covering the angles, depth of field, focal lengths/angles of view, lighting, etc., etc to maximize his creative choices for the final cut from lots of footage of the same scenes. Very smart (though laborious). The goal is the final product - not bragging rights about how many cameras you used, or didn't, or how many takes it took. Craft is about outcomes, both creative and financial, not how you got there.
20,000,000 feet = 6,100,000 meter
1 frame = 0.019 meter at 4 perf (considering cinema scope 2.4 here, with 1:1.2 anamorphotics)
6,100,000 meter / 0.019 meter/frame = 321,052,632 frames
321,052,632 frames / (48 frames/second * 2 eyes/frame * 2 exposures/frame) =
321,052,632 frames / 192 frames/second = 1,672,149 seconds
1,672,149 seconds / 60 seconds/minute = 27869,15 minutes
27869,15 minutes / 60 minutes/hour = 464,49 hours
464,49 hours / 24 hours/day = 19,35 days
1 frame = 5120x * 2132y pixels (assuming 2.4 cinema scope here) * 16 bits per color * RGB = 65,495,040 bytes per frame
321,052,632 frames * 65,495,040 bytes per frame = 21,027,354,974,945,280 bytes
Thats 21 peta 27 tera 354 giga 974 mega 945 thousand and 280 bytes when stored uncompressed...
OK, no one will do that. They will make selected takes, convert only footage they need, may rely on B44A 4:1 EXR compression. So we don't really now how large their storage requirement will really be (except they tell us).
Interesting would be which recording format they really use (RED compression settings) and how they proceeed from there:
Do they have hellfast playback using some custom made 6 rockets in a box stereo playback system?
We all heard of that being possible, but has anyone ever seen it?
Do they just transcode to a proxy format?
If yes, how and to which format?
We currently use a CineForm based workflow (using our Converter application with network transcoding) based on CineForm DPC (aka DPX-C) single frame sequences. We get usually about 25% larger than the original RED file, but with RGB not RAW from then on in post. So the real disk space amount of the used clip data and all forthcoming VFX layers will be dramatically smaller than the above scenario would imply.
Coming back to the equation in terms of disk price:
I guess they using something like the latest DataDirectNetworks SFA12K that puts around 6.7 Petabyte in just 2 * 19" Racks (comparably small!) and with a target speed of 40 GByte/s???
That system won't come in cheap as well, but surely cheaper than the film development - and reusable for some time.
Not to forget about the pile of LTO5 tapes to back this all up and safe it for a little time.
A real shame is that printing to film still gives you the longest living and most secure archive...
|« Previous Thread | Next Thread »|