This thread is like phone sex for camera geeks, it's a bit cruel really
Seriously though, a big thanks for taking the time to post on your progress.
Gripping stuff, thanks boys, keep it up!
With the help of all my partners, I have been heading up the end-to-end workflow on this film for about 3 months, along with a slew of other Epic 3D and 2D projects. I will comment on this workflow as best I can in the coming weeks. For now, I've included a picture of our OUTPOST carts which are now optimized for Epic downloading, processing, and delivery to AVID, PIX, and iPads for editorial and dailies viewing.
What is important for all soon-to-be Epic users is to start brainstorming a way to manage their Epic data which, compared to RED MX:
1. is 2-3 times larger than MX RC42
2. is saved to cards that are 128GBs (8x larger than 16GB CF)
3. requires window-burned 3D side-by-side file creation (for 3D shows)
4. requires convergence adjustments (popular 3D trend is to shoot parallel)
5. requires 3x copies of all media prior to leaving set each day
At an average of 6GBs-8GBs per minute, what many people will run into on their first Epic adventures is the potential (and dangerous) bottle-neck of improved image fidelity in the form of increased file sizes. This needs to be well thought out on Epic productions to insure there is no bottle-neck in the download and processing stations. Without a newly optimized set of gear, productions could fall behind set and the data-techs/DITs/wranglers will be left on set backing up and processing for hours after wrap. A surge in increased processing times could have a negative impact on the Epic if producers notice a significant delay in the ability to provide files in the same manner as they are used to with MX.
On shows that shoot to uncompressed DPX files (such as the file-based adaptations of F35, Alexa, and Genesis), many producers are holding transportation longer than anticipated due to the delay of the data. I'd like to encourage people to start benchmarking their systems now and upgrade where necessary to adequately handle the size of Epic data which is similar to the size of uncompressed HD (upto 180MB/s), which can be 10GBs per minute. The simplest way to start preparing for this is to make sure you can maintain the ability to move files in real time. In other words, your system needs to be capable of moving up to 8GBs-10GBs to multiple locations in 1 minute's time. RED has helped solve the problem with the new RED Stations, which deliver incredible speeds of SSD transfers to multiple targets. From there, you need to optimize your systems with the ability to:
1. check sum all data to each target directory (2-3 total)
2. color, sync, converge every shot
3. process AVID, ProRes, H264 (we do all 3 for every mag)
4. transfer to a shuttle drive
Some hints to begin designing your workflow (both on set and in post) is to:
1. 100% eliminate dependance on FireWire800. Reliance on even 1 lane of FW800 will likely start putting Epic shows behind schedule well before lunch.
2. Rockets are standard. Accelerating your viewing and processing has always been a must, but with files of this size, reliance on CPU power alone to debayer in a reasonable amount of time is simply impossible.
3. Look at different transfer potentials. We have been relying heavily on SAS (we did all of Pirates 4 with MAXX Digital SAS Raids) and lots of eSata shuttles. GTech now makes an incredible, quiet, and mobile SAS Raid (GSpeed eS Pro)which can easily deliver over 500MB/s R&W (4-8TBs @ Raid 5). In other words, you don't want to get stuck moving multiple terabytes per day over FW800.
4. Multiple Computer Terminals. With 2+ camera shows (or in 3D 4+ camera shows), you will need to have multiple computers to aid in the processing, checking, viewing, coloring, transferring, etc, of these files. Design your stations to handle more than 1 CPU.
Granted, all of this is brand new, and this film has taken a lot of the arrows, which means we can share our experiential knowledge with the community. Deanan and Matt have done incredible work with REDCine-X. (I think we might have received 10 builds inside of a week). REDCine-X is a critical part of the process, and with the addition of a number of new tools in the app, Epic shows are poised to out-perform the competition with the power RC-X delivers to on-set operators.
Lastly, the most important tool for a solid workflow are the genetics of its team. Commenting on the on-set workflow, I have to give major props to the small, but powerful team we've assembled for this Epic production.
1. Files come from the camera through Brook Willard and head to what I call his "lilly-pad" station. Brook colors each clip (LE & RE) with John, sets convergence with Rob, and creates an RMD profile which independently accompanies the R3Ds.
2. R3Ds and RMDs are delivered to the OUTPOST stations where the FreeHill team, Steve Freebairn and Zach Hilton, begin the processing tasks. They check sum 2 copies to 2 raids, apply the RMDs, create side-by-side media, window burn it, and create AVID, ProRes, and H264 across 2 terminals and transfer it the eS Pros for delivery directly to editorial.
3. A fully-completed set of synced, graded, watermarked file dailies is processed and files are transferred to iPads for production to take home with them (iPads rarely contain the last clips of the day since principles leave soon after the last take).
I have to say considering all the unknowns of a contract like this, I speak for a number of people on the team who have said we are happy with the results thus far. With the right amount of pre-production and testing, every Epic job can be as smooth as a RED MX - only the stakes have gone up, so we all need to re-evaluate our talent and tools to insure we deliver the best results to all our clients.
Mike that was one dense yet incredibly informative post.
It's the reason why come back to reduser.net throughout the day.
Excuse me while I digest everything you just said but in the meantime
thanks for that post.
Puts a lot in perspective.
|« Previous Thread | Next Thread »|