Hey guys, just wondering what your 4k workflow is currently. I have a 2013 iMac 27 i7 and early 2018 MacBook Pro
Im looking for any suggestions on how to improve my own workflow until I can purchase an iMac Pro or new Mac pro
Currently, I film 4k on 3 x gh5's plus a Mavic 2 pro (normally 24fps)) and import directly into fcpx (but I dont use proxy although im changing this as my machines are starting to lag badly. (all libraries are stored and worked from an external Lacie 6tb raid. I had thought about buying a 1tb SSD and work on my skates project or sometimes I have a few libraries open (does this slow fcpx down?)
I edit on a 1080 timeline , use various effects and plugins, Cinema Grade to colour and add LUTs if needed then export via the share menu. In an earlier post I noticed someone saying they use a 4k timeline then use compressor to share to a 1080p formate (is that better). Then transfer the library to a different drive for storage
Any suggestions to improve this? Im open to all ideas to speed up my workflow and beachball spinning
Your RAID should be fine, I'd not change that. Is it USB 3 or T'bolt? The one thing that would significantly speed up your workflow is an upgraded GPU, which you can't do. Keep your system drive clean as possible, not much else you can do, realistically.
As far as timeline dimensions, edit what you're going to export. Plus the 1080 timeline lets you zoom/pan your 4K clips if that ever becomes necessary or useful. Editing on a 4K timeline I don't think will buy you much. I'm on a 2015 rMBP and an OWC RAID, edit 4K. Not the best, but not horrible. I do only keep one library open at a time. Two if I'm using my Stock Media library. Quit and relaunch FCPX once or twice a day seems to help.
gaddster wrote: ... I have a 2013 iMac 27 i7 and early 2018 MacBook Pro
Im looking for any suggestions on how to improve my own workflow until I can purchase an iMac Pro or new Mac pro...
Currently, I film 4k on 3 x gh5's plus a Mavic 2 pro... import directly into fcpx (but I dont use proxy although im changing this as my machines are starting to lag badly. (all libraries are stored and worked from an external Lacie 6tb raid. I had thought about buying a 1tb SSD and work on my skates project or sometimes I have a few libraries open (does this slow fcpx down?)...I edit on a 1080 timeline , use various effects and plugins, Cinema Grade to colour and add LUTs if needed then export via the share menu. In an earlier post I noticed someone saying they use a 4k timeline then use compressor to share to a 1080p formate (is that better). Then transfer the library to a different drive for storage...
Just editing 4k H264 is predominately a CPU issue not I/O or GPU. Once you add effects it can become GPU bound.
The I/O rate of 4k H264 is modest, typically about 12-15 megabytes/sec per stream. Most disk drives can handle that.
Using proxies greatly lessens the CPU load since each proxy frame has 1/4 the pixels and ProRes is all-intraframe -- IOW easy to decode.
I have a 10-core Vega64 iMac Pro and it's somewhat sluggish on certain 4k H264 codecs, esp. Sony XAVC-S and DJI's. So an iMac Pro won't make editing 4k H264 blistering fast. The 2017 i7 iMac 27 is actually more responsive since the Kaby Lake version of Quick Sync is faster than AMD's hardware acceleration which the iMP uses.
The bottom line is you often need proxies if you want fast, fluid editing -- even on an iMac Pro.
The other solution is shoot ProRes. My doc team has several cameras inc'l GH5 which we capture via HDMI to an Atomos Ninja V. That 4k material can easily be edited without proxies, but it's about 6x the size of H264.
At 6x the size, I/O during editing becomes a more critical issue for ProRes. Each 4k PR 422 stream can require 70-80 MB/sec.
For our Inspire 2 we have the ProRes option on the X5S camera and that makes a big difference. At 10-bits and ProRes 422HQ or 4444XQ, the quality and editing latitude is vastly better than H264. However that's not available for the Mavic or Phantom. BTW if shooting D-Log I strongly recommend not using DJI's H265 implementation, only H264. We saw significant issues with banding and posterization on the Inspire X5S when shooting H265 D-Log, even though it has the same bit rate as H264. In theory H265 at the same bit rate should have better image quality but it's worse.
1. create folders with subfolders to drop the camera footage and assets in
2. creat Library, import folders and structiures and create proxy files at this point?...
The right workflow starts before shooting. Make sure all cameras and audio recorders have the correct time of day. Otherwise it can be difficult to group clips together to sync a multicam (which includes single camera with external audio).
Have a top-level folder with the project name, e.g, Housing Documentary. Within that have a folder with each day's shooting, and within that a folder for each camera. On large projects you might have a subfolder per shooting location and within that folders for production day and below that camera name. If multiple operators are used, combine the operator name and camera name, e.g, JoeDVX200. Otherwise it can be difficult to tell apart material shot by two people using cameras which share identical codecs, e.g, Sony A6500, A7R2, A7R3 or Panasonic G85, GH4 and GH5.
Ideally rename all files after offload to avoid duplicate filenames. Some cameras like the Sony Alpha series give no in-camera control over video filename and start over at C0001 each card. FCPX internally appends a "uniquifier" suffix (fcp1), (fcp2), etc, but this does not prevent an XML bug which can cause spurious duplicate clips.
We find the best practice is append a 5-digit incrementing serial number to each file. It's also easier to manage clips in post (esp. in a collaborative environment) using a 5-digit serial number. There are lots of renaming tools and even Finder can do simple renames. However we use the 3rd party tool "A Better Finder Rename":
After that ensure FCPX preferences are set to create keywords from Finder tags and folder names. Upon import this automatically keywords material from each operator's camera. This is needed to later label each batch of clips from a given camera with a camera name or angle name for multicam sync. Our DIT sometimes adds metadata via Finder tags, and this also becomes keywords.
Import with "leave files in place", but do not import bare AVCHD files this way. Those should be imported within the library or externally re-wrapped with EditReady before import, else it will cause major I/O performance problems.
To maintain a "lean library", it's best to set library properties to store the cache in a separate folder. That enables easy backup of multiple library copies in Finder by right-click on the library and picking "duplicate".
By default proxies go inside the library. This also makes the library too large to easily make multiple backup copies of. To keep the library small you use the Inspector to set library storage properties to some external folder location. Media imported with "leave files in place" doesn't go there but proxies will be placed there, keeping the library small.
After library storage properties are set, you can create proxies as part of the import process -- there's an FCPX preference for this -- or afterward by selecting the clips in the Browser, right-click and pick "transcode media". Do not pick Optimized Media unless you have a specific reason. Usually proxies are all you need and they are much smaller.
When external proxies are created, the disk volume name is stored in the library. This cannot be changed, so all subsequent use of those proxies must be on a hard drive with that name. Proxies (unlike regular media) cannot be relinked.
again thank you for such a detailed response, im going to have to read that a couple of times and make notes.
Historically I just import the files in then, assign the camera angles with a name etc, the use keywords to organise. Ive not used proxy as previous stated but thats about to change as I need to edit faster