When broadcast transmission deadlines are tight, you need an NLE that's fast and robust. John Matthews describes his Final Cut Pro X workflow that enables him to deliver quick turnaround shows to the BBC with ease.
As part of BBC Scotland’s public service remit each year, we have produced 2 “highlights” programmes reporting on Events at the General Assembly of the Church of Scotland. It’s a lovely job in the bustling City of Edinburgh. Edinburgh has been rated the top “cultural and creative city” of its size in Europe. It’s a fantastic place to be.
I just looked it up and I think this is the 10th one of these jobs I have done. Based in this imposing building on top of “The Mound”, alongside Edinburgh Castle. This has become my home for a week in May.
The Event runs for 7 days. We usually include highlights from the floor of the debates, augmented with a couple of films, on current topics. The BBC schedule for transmission means that the first show goes out only after the first day of coverage, giving us a bit of a problem as to what to include! Hence the pre-cut inserts.
This year, youth was the theme, so a couple of weeks earlier, a crew and reporter travelled to Jaffa in Israel, in the old section of Tel Aviv. It was a fairly straightforward piece about young kids in a school there, many of whom speak 7 languages, contrasted with an ethnically diverse and quite poor school, in Edinburgh, where English isn't the only language spoken. The kids communicated by writing letters and sending video diaries.
All the insert material for Programme 1, was shot log at 1080 25p, S-log3 on the FS7 and edited at home on FCPX beforehand.
I am rarely at the shoots and so usually have little control over how the camera originals are handled. They tend to arrive already on a drive, and I’ve never quite sure what to do at this stage.
I used to always make Card Archives and import those, but it seems to be needless duplication, if the cards have already been wrangled. So now I import directly. All media is stored in a folder external to the Library, so that 2 of us can collaborate down the line.
I use Chronosync to back up the whole of the media folder to another drive, running a schedule every hour. Each camera card arrives in a folder on the source wrangled drive. The folder name becomes the Camera Name, and we create Smart Collections to contain those.
First hitch- X applies a default LUT (Sony S-Log 3) at source, but gets it wrong and blows out the highlights. This seems to be a problem with this camera and having the LUT at the front end, if it’s severe, above 120%, then the detail becomes irretrievable.
Default LUT applied
No LUT Applied
FCP has identified the clip as being REC 709, but allowed the levels to go way over. Ideally I’d like this not to happen, (maybe the LUTS are for REC 2020?) as you can see from the original log, there is plenty of latitude. One solution would be to reduce this LUT and save as a custom one, with enough headroom for further tweaking. Then you could have X apply this at source, without fear of losing detail, but this isn’t an available feature.
As a workaround, I found a Custom LUT online that had less overall gain and applied that at source. It’s not ideal, because there is very little time to fiddle, any grading really should be to augment an already usable image.
My first pass is to identify what's what, using a simple naming convention: IV - Interview, CA - Cutaway, PTC - Piece to Camera etc. I “Batch apply” these to the "Scene" fields for each whole clip type designation. Then “Batch Rename” all the clips, using the built in features in FCP, to retain the original clip name and append with these codes.
Smart Collections are again created to contain them. The camera metadata is often wrong on these jobs, so now would be the time to also add a custom name at the beginning.
I like to make the distinction between Clip metadata, which is not Range-based, eg. Scene Take etc., and specific editorial content, which is Range-based and so Keyworded. I do use Keywords for whole clips, but try and keep them to a minimum. The Smart Collections are “smarter” and need less intervention once set up.
So Smart Collections to house types of material, Folders for Keywords. Basically “Places”- Where are we? “People”- who is it? “Other”- what are they talking about. Folders are generally a great visual aid and allow for viewing of the entire contents if clicked on.
I find it easier to keep Sequences (Projects) in their own Event. When there is only one deliverable like this, I make a dated Smart Collection to contain the Current edit. This is the first Sequence I make and ends up as the Master. It’s simple enough to Duplicate or Snapshot edits at timely intervals along the way
I’ve found that lots of Projects really bloats a Library and Snapshots can and do cause duplicate Compound and Multicam clips, so I use sparingly and only for revisions not Backups.
The Best bet for backups is to use this fantastic App. “Backups for Final Cut Pro”, available on the App store, for next to nothing. I have it set to Backup every 5 minutes, which is preferable to the 15 minute built in ones. The backups are incremental and you get a notification when they are happening. No one should be without this.
I used SpeedScriber for transcriptions. Keyworded all the contributors and broke into Acts, also with Keywords. Separate Events for Rushes, Sequences, Music, Graphics and Voice Over. I’ve tried pushing the transcript back to FCP, but it currently isn’t very user friendly to view the text, so prefer to use SS as a standalone app pointing to the media. The playback engine is amazing and it’s so quick to find mentions of words.
I gave Roles to the clips as follows PTC, IV, plus various Music Roles. That meant anything not re-designated, was by default Sync FX, so I just kept that as Dialogue. I’m handling the final audio so I just keep it as simple as possible.
Reject and Rate with Favourites in the standard way.
I try to remember to pre-select correct audio channels and roughly set levels before editing. I'm a big fan of always keeping audio with the clips, regardless of whether it’s required. Is there a button that could de-select audio in the Inspector to save time with this? It's that 3rd missing edit option "Edit with audio, but de-selected”
With thanks to Alex Gollner, I’ve learnt that you can turn off channels in the Inspector by Pressing V as you skim. An alternative method. Basically I don’t see why you would ever not want the audio inside the clip. Even when sending to a Dubbing session X2Pro has options for ignoring disabled or inactive clips.
Using a Sony OLED PVM A250 colour monitor, I managed to grade without removing the custom LUT I'd applied at source. The Grading tools in X are more than enough for most things I'd ever do.
The audio was pretty awful, as is the way with single operator shoots and no sound guy. Lot of background noise. I used some basic AU compression and EQ and thought I'd Compound the the whole Project at the end and apply RX7 Denoise and De-Reverb to the stems. The thinking being, that this is processor intense and so can easily be removed, or de-selected and managed at the end of the chain.
However, 2 of us were assembling the show, which was half an hour long, so the cut got passed around a bit. When it returned to me, via a Transfer Library, I found the Compound clip of the Jaffa piece was black in both the Transfer Library and mine. Nothing I could do would bring this back.
Reluctantly, as under pressure, I broke apart the Compound, but this of course lost all my "bus" audio effects and slight audio balancing tweaks. So bit of a gotcha here! Besides you can't monitor the audio properly, whilst inside the Compound and the collaboration in X is still so messy, that I just wouldn't go there, unless things change. I hope we get improvements on audio and collaboration soon from Apple.
The “Lanes View” though, is superb and once you get used to focussing on the area you are working, it's easy, even without the Compound clips workflow. Important to order things sensibly on the left. Batch applying means either searching for that Role as text, or dragging a lasso over portions of a Lane. The latter is quicker.
In the past I'd found Izotope RX7 very crashy, but not so now and my iMac seemed fine with lots of it being applied at clip level.
Although it's a bit of a workaround, next time I would edit all my talking heads inside a Multicam, then at the end export the audio only, from the selections used, treat externally, re-import as AIFFs, create a new angle in the multi and waveform and match them back in. Then switch to that Role! Much less taxing on the system.
Synchronised clips wouldn't give you this flexibility. Apple, it would be so handy to have Timecode as an option in our audio exported files! I’m thinking mulitcams for interviews is a standard workflow, as it gives you so much flexibility even if you only have one angle!
Once on location, it’s basically a 2017 top spec iMac with 40GB of Ram a 12TB Pegasus R6, connected by Thunderbolt. Then a 2015 MacBook Pro, with Cinema Display, and another 2012 MacBook Pro, to do recordings, all through a Netgear Gigabit switch. Not elegant but it works fine at 1080i.
We are supplied with 2 X SDI coax 1080 50i feeds, with embedded audio. We make 2 x Main and Isolated recordings to 2 striped 8TB drives. For resilience, the recordings are done to a separate Mac. The 2012 is perfectly adequate for this. We record every day for 8 hours at Prores LT. Only a fraction of this material makes it into the show.
The networking arrangements are the weakest link in the chain. Really we need a server and I would love a Jellyfish! The bandwidth is adequate, but because FCPX’s media management is so primitive, we either end up with huge files across a network, or we chunk our recordings down to 10 minute slices.
We also multicam the dual feeds and that produces even more media management headaches. We don’t particularly need “growing files” so the 10 minute chunks are the best compromise, until we find a way of flattening multicams. Additionally the 10 minute chunks always give you the option of only pulling in the media you think you are going to need.
The week trundles on and we record various “Piece To Cameras”, on location and “Voice Over” using the punch in tool. I use a Coles Lip mic for this now. Many have reservations, (mainly hygiene?), but it gives complete consistency, incredible isolation and it’s a great prop, that ensures you always get a performance out of the talent!
We use Folders and Smart Collections for Single Camera Acquisition. I find it useful to have a dedicated “Import Event”, for newly imported material. Naming conventions as above if there’s time. The Import Event acts as a “holding area” and saves you hunting for material, when under pressure, until you get it tagged correctly and can be moved. I think it’s important to use the “where” and the “what” with media as appropriate. We have an uncanny ability in our heads to remember where we put things!
Sometimes that is still useful in-spite of the immense power of the FCPX database.
Now, collaboration. Where to start? Well first of all, you could not do this job on Avid MC, without a barrow-load of kit. You would certainly have to use external AMA (Avid Media Access) media, unless you had Bin locking software, which is expensive, (although it is now offered to third parties and there are some alternative solutions). But doing it on Avid would just be such a drudge.
However one thing you can do on Avid, is inhabit the same Project. When I say Project I mean Library in FCPX. This is super convenient, because the metadata flows, you isolate the bits, as required with Bin locking and everything stays organised. With FCPX your Library is your own and so are your clips. This can create headaches when you are under pressure, as extreme care has to be taken to avoid duplicates. Frankly whatever we do it always ends up a bit of a mess.
The rule is XML for simple cuts and a Transfer Library for passing complex Timelines around, but as above, expect the unexpected! Move instead of Copy is the watchword for avoiding duplicates. I usually get the Library into reasonable shape and copy it for the other party and that becomes their starting point, from that point on metadata flow becomes very limited.
Another couple of things, by the way, I’ve noticed, for those of us stuck with interlaced delivery. The features were all shot 25p, but the Live Event coverage is 25i (50i). Delivery 1080 50i- (All broadcast TV I know of is interlaced as it arrives in the home. One other feature had a mixture of both.
So if you are working in a 25p timeline and use 25i material, X will drop a field. When you subsequently cut this material back into a 25i Sequence, X interprets the fields the wrong way round. Basically the flag is set to "none' and so it gets it wrong- in this case “lower field first”. Similarly if you apply any optical flow to 25p material, that ends up being in a 25i Sequence, it will look silky smooth, (it's adding field motion that wasn't originally present) and completely at odds with the filmic look of the rest of the piece. Anyone know a fix for this?- 25p Compound clip maybe? Didn't work for me.
We require rolling credits. The Scrolling Titles Browser Template is still broken, and mostly doesn't produce fields (although strangely, sometimes it does! A 25p rolling title is unacceptably jerky.) I've long since given up with this, so the workaround, to get a smooth roller for TV, is to build as normal, then cut and paste the Title into a 50p timeline, export PR 4444 with alpha, reimport and apply to the background in the 50i Sequence. It will then be as smooth as any interlaced Title from Avid's Title Tool! I'm so glad this works- hate having to go to Avid to do a roller!
The Used Media / Show Unused Media filtering is so powerful in X. I think next time though I would hit the Reject button harder, basically because the Used feature is range-based and not clip based. So with hand-held coverage especially, you still see all the shards of unused content either side, when you filter like this. Could there be a way to only show whole clips that haven't been used?
Gripes are small but worth noting, mostly audio. Audio clicks require manual fades and trimming. A subframe fade preset, sometimes works, but we really need to be able to overlap adjacent similar audio for quick smoothing. "Add fades with handles" would be good. I still find the Multicam cuts produce clicks unless at a zero crossing point.
It's still easy to unintentionally end up with media "Left in Place". Consolidate doesn't always work at Library level. I'd like the ability to walk away with a drive that only has used media and handles, for emergencies, if I have to do any repairs after AQC (Tech Review). You can make a self-contained Library and drag off to another drive. But the process is hampered by having to use Worx4X, or copy much more media than you really need. You do not want to be 50 miles away from your edit station, with missing media. The 10 minute chunks can be a lifeline here.
The finished shows are output Rec 709 legalised and R128 loudness compliant. I just balance the audio on a decent pair of monitor loudspeakers (Neumann KH120s), Export the mix through a Loudness App from Video Toolshed. I will make an offset adjustment if required to be -23 Lufs dB.
We have to make a DPP compliant AS11 file, which is AVC Intra 100. FCP can do this, but you currently can’t edit the metadata, There are expensive apps that allow you to re-write the metadata shim inside the MXF file, but for now I’m having to use Media Composer to do this, which involves an extra video compression step. I’m looking at Resolve, but it will be much the same. Avid MC adds a “Spanned Marker” - a Chapter Marker in FCPX. All the metadata fields for AS11 are available in FCP, but currently read only.
Once the show is complete and has transmitted I delete any generated media and use Worx4X to slim down my Library, keeping only the Sequence clips plus half second handles. This is a reduction of something like 500GB down to 50GB, so Worx4X pays for itself in one operation.
Delivery to the BBC varies. Programme 1 was tight and exempt from any Technical Review, so a ProRes mov is fine and this can be linked to in Avid Consolidated and sent directly to Playback Airspeed Server. (mixed down to XDCam 50).
The second show full AS11 DPP compliant. AVC intra 100 MXF. Nothing ground breaking here, but some quite good process information from acquisition to delivery, that others hopefully, will find useful.