Jump to content

BMCC 4K Feature Workflow


Raj Kowoski

Recommended Posts

Guys:

Planning a feature shoot in 2 months and need some advise from seasoned editors. Here I go:

 

Last spring we shot 3 days in NYC on BMCC 4K with a goal to create spec trailer. Taking this opportunity to learn editing, I got myself MacBook Pro 16GB with graphic card + FCP X. Great experience and invaluable lessons learnt that I plan to carry to this next venture.

 

Feature is probably 90 minute long compared with 2.32 min of trailer. While Mac was able to handle 4K Prores voulme for a small trailer, not sure if it has the juice to handle 90 minute of 4K. May end up shooting some footage on RED Scarlett as well. Question:

 

Can someone please point to an optimum post workflow?

Link to comment
Share on other sites

Are you planning to do 4K Raw or Prores?

 

I would use DaVinchi Resolve rather the doing everything in FCP, as theres a decent free version or the BM cameras come with a full version and its good for making proxy files, reconforming and much better then FCP for colour correction

 

For either format my approach would be to:

1) Transcode the footage in Davinchi Resolve to Prores LT or Proxy HD

2) Edit Prores Proxy files in FCPX

3) when edit is complete export XML from FCPX

4) Load XML into Resolve - conform with original full res footage and grade in resolve

 

There are plenty of online tutorials explaining the process of round tripping between Resolve and FCPX in more detail

Link to comment
Share on other sites

Thanks for the workflow and apologies for being noob. We are shooting 4K Prores 422.

 

In layman's term,

1. drop (or import) Prores 422 files in DaVinci,

2. convert (transcode) this footage to Prores Proxy files (these files will continue to have .mov extenstion),

3. import these lower quality files in FCPX,

4. edit in FCPX,

5. export xml from FCPX

6. import this xml to DaVinci

7. DaVinci is smart enough to line up these Proxy files back to original ProRes timecode source,

8. grade the ProRes 422 in Davinci,

9. export the final product.

 

Couple of follow up questions:

1. How does Davinci align elements such as Titles, Transitions, sound after I import. Will there be any out-of-sync sound issue?

2. Can you please point me to any youtube tutorials if you have those?

This will be a 90 min feature. Prores 422 records at 471 Mb/s and 212 GB/hour. So, I'm looking at around 300+GB of final file. Does MacBook pro 16GB RAM have that kind of power to render?

3. The goal is to apply film emulation to get me that 35mm Kodak Stock look of 1980's horror films. Skin tone, grain, saturation etc. With so many third party film emulation tools in market and this probably will make non-theatrical markets (VOD, DVD), will there be a noticable drop in picture quality if I stay with Prores LT instead of 422? To me, final sound mix is the most important element. Images are a luxury. No offense to DPs on this board.

Link to comment
Share on other sites

  • Premium Member

You need to be quite careful about reliable backup for your footage. By the time you've got any transcoded versions you need, intermediate and final outputs, you're likely to need quite a lot more than I think you've probably planned for.

 

Making more duplicates on hard disk is fine enough, but you are likely to need something better - look into LTO tape.

 

P

Link to comment
Share on other sites

Hi Raj

 

You'd want to re-do the titles in DaVinchi, it has a title tool or you can import graphic titles. Titles produced in your FCPX offline edits are likley to be HD and I guess you want to work in 4K for the grade. I would just use placeholder titles in the FCPX edit for timing purposes. Most transitions can be re-created in DaVinchi - some will be pulled across in the XML and others you'll have to remake in DaVinchi using the internal tools. For complex titles or transitions you might resort to after effects or similar anyway.

 

For sound you can import audio into DaVinchi and make your final deliverables from there. I'm not sure if it handles multichannel audio - but you can at least add a stereo mix. For a feature I'm guessing you might do the mix in pro-tools or similar rather then FCP anyways. The audio should just line up as your exported sound file "should" be the same duration as your video project and plonking it on the DaVinchi - timeline will do it (you can of course nudge it about to tweek sync if needed).

 

I would still used ProRes HQ for the master export over LT - although on a monitor you won't really be-able to see the difference. For VOD/DVD/Broadcast delivery its going to be compressed down to a very small file and its always best to start with the highest quality original. If your working at 4k your going to need something in the order 15-20+ terabytes of space for the camera rushes, backups and transcodes if you've got a reasonable shooting ratio. So the difference between HQ and LT for the master file is not going to make a huge difference to your overall storage needs.

 

Your laptop should be able to cope with the project, it might not be particually fast when working at 4k. Having lots of fast storage space is going to be key. One thing you could do is work in sections - e.g cut it as 20min reels, so you never have a massive timeline open at any one time.

 

I have a BM 4K at work and all the projects I've done on it so far have just been shot in HD Prores, because the 4k data rates have been unmanageable and too expensive in storage. The in camera HD has been good enough for my purposes as 4K distribution is pretty rare

Link to comment
Share on other sites

I strongly advise you NOT to edit off your originals in 4K. Keep them real safe the very moment you've shot them. A double backup to LTO-5 tapes that are then kept at different places is the cheapest and safest option. It should be reasonably cheap to just hire an LTO deck off any IT rental house during the shoot. If it's all to be on one location it shouldn't be too difficult to setup a place in the location as the lab where all the archival happens. Have a look at either Silverstack or YoYottaID for the entire ingest and archival process. They might seem expensive, but there's nothing more expensive than failing to archive your data. Actually, I think that Silverstack can both ingest, archive to LTO and generate the proxies with a LUT baked in one single step.

 

Instead, transcode everything to prores 422 proxy (if editing on FCP) and work off any drive that's large enough to hold the transcodes. Also, if transcoding on Resolve, make sure you tick "assist conform using reelname --> use filename as tape id" on the project's settings window. This will guarantee that you have at least two bits of information on every transcoded clip pointing to the original rushes. You'll have both the clipname, that should remain the same, and the reel id in the transcoded clip's metadata.

 

Otherwise, you're looking at a huge and unnecessary investment on an external raid just to get a picture lock and the need for a beefed up station for a smooth edit.

Also, test the workflow before shooting. If in doubt as to how the process goes, do consider hiring a professional DIT if it's only to setup and explain the workflow before the shoot happens.

Trust me, you don't want this stage of the process to become a nightmare. Cover this base quickly and let the actual filmmaking happen without headaches.

Link to comment
Share on other sites

  • Premium Member

Final Cut X is a toy, it's extremely limited in capability. However, if you haven't edited before, it's probably the easiest tool to learn. When working on a feature or something that requires lots of media management, I would highly suggest taking a course and learning Avid because the tool's inside, allow for much better clip finding and organization.

 

I spend most of my time cutting blackmagic material, so I have a good workflow which seems to make sense.

 

Buy a bunch of cheap USB3 2.5" portable drives. The max you can get today is 2tb. These drives are very robust and because they don't have firewire bridges or external/internal power supplies, they're almost impervious to the type of failures you get with other drives. Not fast enough for editing, but absolutely fast enough for storage of original media.

 

Each one of those USB3 drives will contain your camera original. You can put as many complete "reels" as possible onto each drive. Each reel should be put in a folder and you should alter the clip name to reflect the camera reel. So if the clip name is BM_Date_clip name. It should just be reel name and clip number; A105_12345. Then you will duplicate them so you have two drive sets an A and a B which are identical camera originals. The A set will be stored in a safe locally and the B set should be stored at another location just incase.

 

Before you store them, you will want to transcode the material from 4k down to something more usable like 192x1080, which is standard HD for broadcast. I'd use the standard Pro Res 422 codec, as it's not over-fat like the HQ codec and it's not missing data like the proxy or lite codec's. With Final Cut X you can use compressor to do this without over complicating things.

 

You will transcode everything to an external plug in the wall 3.5" 7200 RPM thunderbolt drive. (since your laptop is new). Once transcoded, you will duplicate the transcoded media onto an identical drive.

 

So before you start editing, you will have 4 copies of your media.

 

2 copies of the camera originals

2 copies of the editing/transcoded files

 

I always store one of my transcoded drives in my safe until the job is finished. I also update that drive with any changes reflected on the first drive. It's important to use these two drives together as a backup method for your project sequences as well. This way if something went catastrophic with your edit bay, you'd have all your edits. A lot of people export EDL's daily, I generally duplicate my project's every few days.

 

When you edit, you will be working with original media… I know it's not, but you can call it your original. The "CAMERA REELS" are never to be touched. From your seat in the edit bay, what your cutting will be your final unless for some reason you need to upres back to 4K for DCP/theatrical. You will color using the same media as well.

 

I would absolutely color with DaVinci, but it's tricky to learn. I would try to take a course if you can because the ramp up can be time consuming. Great if you have nothing but time, however if you have a deadline, I'd study and learn before touching. I know FCPX has color tools, but they're garbage compared to DaVinci. Apple screwed the pooch after FCP7 and hasn't fixed it yet, though they're getting closer making FCPX work.

 

I hope some of that resonates. :D

Link to comment
Share on other sites

Tyler, the cost per GB of hard drive vs LTO tape is something like 4 to 1. There's also a big difference in speed, 250 mb/s for LTO vs the 90 mb/s for the cheapest external usb 3; and a massive difference in the reliability of said media. LTO is rated for decades.

The standard bitrate for offline compression, Avid for example, is 36 mbps. This is Avid DNxHD 36 8 bits. Prores LT is the closest equivalent. It's possible to go even lower, but then you start asking the processor for extra cycles to decompress the media purely for playback. The 36 mbps number strikes the best balance between viewing quality (perfectly fine for an intermediate codec not meant for online) and the demands on both the station and the media where the proxies are to reside during the edit.

Link to comment
Share on other sites

  • Premium Member

Tyler, the cost per GB of hard drive vs LTO tape is something like 4 to 1. There's also a big difference in speed, 250 mb/s for LTO vs the 90 mb/s for the cheapest external usb 3; and a massive difference in the reliability of said media. LTO is rated for decades.

I've built many post facilities around the US and we've always used LTO libraries for backing up online storage. When you have over 25TB of spinning disks (multiple raid array's) LTO is a great way to keep that data secure. On the install's I've done, we've averaged around 90Mbps per second on 4gb fiber LTO5 drives, with uncompressed media.

 

From my personal experience, being an editor for two decades… most clients want the camera originals back when your finished editing. Back in the day, we'd just hand them a box of tapes or film. Today, we hand them hard drives, lots and lots of hard drives. LTO isn't a "hand off" medium unfortunately. Most clients will want a plug and play solution which allows them to read your drives without special hardware. This is why I personally do everything on spinning disks. It's a real pain in the ass, but if you do the leg work up front, it actually works well. Today the 2.5" 2TB drives are around $85. If you buy in bulk, you can get the price down a bit. It's much easier to organize in the finder without needing a backup utility as well. This insures your backup disks are perfect because you can read the media before sending them off. It's hard to dump all the media from an LTO onto a disk, read it and verify it all works before sending it off. You honestly don't know if everything on the LTO is working until you do that.

 

So anyway, that's why I made the suggestion. It would be awesome to deliver LTO 6 to all clients, when that day happens, I will be in heaven. :)

 

The standard bitrate for offline compression, Avid for example, is 36 mbps. This is Avid DNxHD 36 8 bits. Prores LT is the closest equivalent. It's possible to go even lower, but then you start asking the processor for extra cycles to decompress the media purely for playback. The 36 mbps number strikes the best balance between viewing quality (perfectly fine for an intermediate codec not meant for online) and the demands on both the station and the media where the proxies are to reside during the edit.

This is a great subject and it's a topic of heated debate amongst editors around the globe. Most of the television guys have excuses for using proxy files, that boils down to storage and multicam. When you have 12 different camera angles to choose from, that's a lot of media sitting on drives. Plus, once you get past 4 cameras, the system will grind to a halt if you aren't using a proxy resolution/bit rate. For those people, proxy based editing is a necessity and those producers/editors are use to working that way. Plus, most of the jobs I've had go to broadcast and they all capped out at DNX115 because that's been determined as the lowest acceptable resolution for broadcast.

 

For anything outside of television or multicam, editing in a resolution acceptable for finishing is critical. The biggest reason why comes down to finishing expenses. You can very easily finish a film at home with DaVinci and Pro Tools and have something tangible in your hand at 1920x1080 so you can show people your film. You wouldn't want to skimp and say "it's a low res proxy version" that doesn't make any sense. You want something you'll be proud of, something you can show people knowing it's what your film looks like. I've done lots of experimenting with Pro Res 220 and 147 and have found them to look nearly identical. For smaller shows for DVD/BluRay and web distribution, I shoot and edit everything in Pro Res 147. For middle budget shows which may see theatrical distribution, I either shoot in 220 or Raw, depending on budget since I can't edit in raw. I've cut features with Pro Res 220 and had no problem with anything. The edit system worked just like it would with proxy, only I knew that what I saw was exactly how the film will look.

 

So yea, I get the proxy concept, but if you aren't running crazy multi-cam and if you're smart with your dailies, it's not a problem to edit in any codec you want.

  • Upvote 1
Link to comment
Share on other sites

  • Premium Member
250 mb/s for LTO

 

John, assuming you're talking about megabytes per second, LTO6 only does 160, and that's on paper. In practice, with files in the tens of gigabytes (mixed duration ProRes quicktimes), via LTFS, on one of the most common half-height drives, I'm seeing 120ish. Source in those instances is an 8-disk RAID10 which is certainly very much faster than the tape, and the CPU is far from saturated, so it's not being throttled. I think that 120MB/sec is extremely respectable for a linear magtape, but it certainly isn't 250.

 

Also, while 2TB disks are about £60 and LTO6 tapes are about £15 in fives, the 4:1 cost advantage is achievable only in the long term. Assuming most users will required the drive and SAS card and that the value is something like £1800, you can store 60TB before you've paid for the hardware, and many small indie features might not actually shoot 60TB (or 30, duplicated).

 

Now, let's be clear: I'm a big fan of LTO and I wish it were more widely used (then it'd probably get cheaper) and I regularly advise people to use it. Hard disks are not a good archival or shipping medium and anything from an even vaguely serious short film or web series upward should be using LTO, but it's as well to be clear about what it can and can't do.

 

Phil

Link to comment
Share on other sites

  • Premium Member

Good thoughts Phil.

 

One point about 2.5" 2TB drives… they're MUCH more reliable then 3.5" drives. For sure not "archival" but LTO isn't really archival either since the tapes are only guaranteed for 25 years.

 

Unfortunately, the only true archival format is film.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...