[Joe Marler]

[Tim Wilson] “…we were mostly talking about speed…The context for our conversation was their work on-set, where they were not just cutting, but roughing out VFX… and audio…Glenn’s feeling is that the speed…of FCPX enabled them to save big money on set, which is where the biggest savings often are. “

Hearing about how FCPX is used in upper-echelon scripted feature films is interesting; it improves the product. It’s fun to discuss. However it is only partially related to the kind of work many of us do.

Those guys are (understandably) very focused on high-end VFX and audio turnovers, on concurrent collaborative editing by a large team, etc. Every second of on-set time is expensive. Their use of FCPX’s keywording and metadata is aligned with a scripted production — references to page numbers, scene numbers, etc.

I would like to see similar workflow stories from a medium or large documentary team. It would be more relevant to those of us doing doc work and even some types of event, investigative news and corporate work. With docs the exact story may not be scripted but is often discovered during the editorial process. There is less urgency about on-set production cost, since much is shot in the field without a big crew. This places different emphasis on how the editing software is used. E.g, shooting ratio is usually higher for docs, program length shorter, more diverse material is incorporated, often lots of stills, etc.

Speed is still important but the various software features may be used differently than a scripted narrative because of the dissimilar production and post-production workflows. E.g, If Ken Burns edited a documentary today using FCPX, what would that workflow look like and how would it differ from Focus?

Have there been any behind-the-scenes accounts of FCPX being used in docs?

[Joe Marler]

[Walter Soyka] “2 GB/s is absurdly fast for most applications. It is, however, less than half the speed of the current state of the art and barely fast enough for a single stream of uncompressed 4K at 60fps.”

According to my calculations, the data rate for DCI 4k @ 60 fps is 759 megabytes/sec. See https://www.theblackandblue.com/2011/03/12/how-to-calculate-red-camera-data-rates/

The HP Z820 workstations used to edit Gone Girl “only” had 1.8 to 2.6 gigabytes/sec I/O bandwidth:
https://www.4kshooters.net/2014/08/13/david-finchers-gone-girl-1st-feature-film-shot-entirely-in-6k/

They probably shot it using 4:1 or 6:1 “Red Raw” compression which would lower the data rate (at 4:1) to about 167 megabytes/sec. If anyone knows otherwise, please correct this.

It was edited essentially using off-line proxies on a 1080p timeline, which greatly reduced the required I/O bandwidth. Of course you ultimately need to conform the footage so at some point you have to process the entire full-resolution locked cut, probably in uncompressed format. For a 2.5 hr film that would be about 6 terabytes (if 6K), however I think the final frame size was 4k.

Most people are not shooting in 6k Red Raw, but the above data rates were sufficient to handle that, and capacities were very roughly in the 2 GB/sec range.

In a more approachable scenario, the Mac Pro that Mark Spencer used on this job could move 2 terabytes within 45 min, or a sustained average rate of 741 megabytes/sec. That is simply using Pegasus RAID arrays, nothing exotic (start at 03:45):
https://www.youtube.com/watch?v=qvl49PvHdZU

[Joe Marler]

[Phil Lowe] “…I hate the trackless, magnetic timeline. I hate the audio controls…I can’t stand skimming in the browser…I don’t like thumbnail view and rarely use it…In short, there’s virtually nothing about the interface or controls I like…Going any further with FCPX, quite frankly, doesn’t interest me in the least.”

You are not the first to feel this way. Mike Matzdorff, assistant director of the Will Smith movie “Focus”, had a very difficult transition from Avid to FCP X. Early on he said “FCPX is utterly unusable”.

His first attempt to use FCPX was editing a little video interview for his wife. He figured it would be a piece of cake for him — a professional Hollywood film editor. He struggled for a long time, gave up and had to finish the video in FCP 7. He was so angry he asked for a refund from Apple.

But he didn’t give up, ultimately edited that feature film using FCPX and wrote a popular book about it:
https://amzn.com/B00UO2NA8I

In this interview, Matzdorff said once he stopped trying to make FCPX work like Avid he made good progress. His described it as “you…have to bend your mind a little bit, and when you do it’s fast and it’s cool and it works”.
https://www.youtube.com/watch?v=pmHwME_sK30

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.

[Joe Marler]

[Steve Connor] “importing 4K XAVC material into FCPX can be VERY slow. It’s worth it however as 4K XAVC in FCPX is much more responsive than in PPro”

When I import UHD 4K XAVC-S from my Sony A7RII, it is definitely not slow. The difference may be importing from the camera card (which does not allow “leave files in place”) vs from a local hard drive copy of those files (which does allow it).

With the camera files on my Pegasus R4 RAID, I just did a test import of 172GB of material. I could drag and drop sequential batches of files to the event, one after the other, as rapidly as I could move the mouse. The delay from dropping each batch on the event until the thumbnails appeared in the Event Browser ready to edit was 1-2 seconds. For a couple of minutes it continues a background task for “generating audio waveforms”, but this doesn’t interfere with editing.

I did the same test on the same machine (a top-spec 2015 iMac 27) using Premiere CC, and it also was pretty fast on the import but each batch caused a several-second “importing…” dialog, during which the UI was locked and you couldn’t do anything else. This has little practical impact — it’s just a few seconds, but from import action to first available edit, FCPX is a bit faster — at least on this material, machine and operating system.

However as you observed, after import on 4K XAVC material, FCPX is much faster and more responsive. Even if you exclude the skimmer performance in the Event Browser as a special case, it’s far more responsive to JKL commands in the timeline. Also the viewer update rate during FF or REV is probably 10x faster. Activity Monitor and iStat Menus shows the CPU levels are much higher in Premiere than FCPX when doing similar operations. So the code path is just not as efficient in this case.

How much difference does this make in producing the final product, IOW “what is editing speed?”. The human editor’s proficiency with the chosen tool probably makes a lot more difference. Specific features of the suite can make a big difference if needed for the workflow. E.g, I’ve seen editors spend 30 min playing around with the FCPX audio EQ and filters when Premiere’s Audition spectral editor could have fixed it much faster.

[Joe Marler]

[Simon Ubsdell] ” FCP X is dog slow when it comes to handling lots of material like this.”

In this case I had 200GB of mixed 1080p material, some of which was AVCHD. FCPX cannot import AVCHD with “leave files in place” so will copy those to the Library. OTOH I have imported about 130GB of H264 4k material from an AG-DVX200 and A7RII in about 1 second.

For some media types (e.g, GoPro, AVCHD and MP4 files from some camcorders) FCPX can go into a “Processing files for import” phase” but this is typically only a minute or so, even for lots of material. Likewise those codecs may produce a “generating audio waveforms” phase which takes a while, but is a background task that doesn’t prevent editing.

Except for these cases, FCPX seems to import extremely quickly. It generally doesn’t strike me as slow to import when using “leave files in place” on a locally-attached RAID array.

Premiere CC likewise does not instantly complete all import tasks for those media types — its “conforming” phase can take a little while. Very roughly I’d say it’s overall about equal to FCPX, although there is some variation from case to case.

[Joe Marler]

[Oliver Peters] ” it still doesn’t provide the human-ordered visual juxtaposition that re-arranging frames in a bin – or images in a lightbox, for that matter – gives you. Which is my point – FCPX is geared around UI norms for computers, but not necessarily humans.”

This is an interesting point. In FCPX you cannot drag clips in the Event Browser to a user-defined order. In Premiere you can, creating a type of “light table” or storyboard. With CC you can even create trimmed subclips, put those in a bin, then in icon view drag them to a preferred visual storyboard order.

It is ironic that FCPX cannot do this since Apple originally strongly promoted a “spatial” UI metaphor. I think that’s why even today you cannot do a range selection in Finder icon view — they are considered spatially arranged so (technically) there is no first and last. In UI design there have been many debates about spatial arrangement vs alternatives. Apple has gradually moved away from this and the Finder behavior is probably vestigial.

The Event Browser in FCPX has various selectable sort orders, just not a user-defined option like Premiere has. It seems like this would be a useful enhancement.

The opposing view is from a relational database standpoint, sort order does not exist. The data is in unordered rows and an order only manifests itself when a query is executed. You query by attribute, not by saying “show me the next item after the previous”. Since FCPX is so database-oriented, maybe this mentality prevailed. However the items in the Browser are more akin to a query result (which can have order) than to the underlying data. I don’t see any reason why they could not be spatially re-ordered by the user, especially since FCPX workflow emphasizes doing as much work in the Even Browser as possible.

[Joe Marler]

[Simon Ubsdell] “I set an IN at the start of the timeline and and OUT at the frame where I wanted to keep something. And pressed delete (Shift X). I set an IN where I wanted that shot to end and then moved to the start of the next shot I wanted to keep and hit OUT and delete. Given the volume of material, I was skimming over a lot of intermediate shots that I was wanting to reject and rejecting them all with a single keystroke”

Maybe I don’t understand your procedure but it appears you are appending one day’s worth of material to a timeline, then you fast forward through the TL setting IN and OUT points and ripple-deleting rejected regions, one by one. Is that right?

Since you have 16 hours of material, you’d be doing this over and over, once for each bin of dailies. Or if the software could handle it I suppose you could append a 16 hr timeline. I tested a 20 hr H264 1080p timeline on FCP X and it ran but was very sluggish. It is definitely not optimized for that.

By contrast when I imported 200GB (20 hr) of that material into FCP X with “leave files in place” this took 5 min and it was ready to skim in the Event Browser and mark ranges. I’d estimate the frame rate during skimming is at least 20 Hz — even on camera-native H264 4k material. This was on a 2015 iMac 27 but it seemed equally fast on a 2013 model.

In your procedure, within a shooting day the order of the non-rejected material in the timeline is based on how that was appended. Since you are working sequentially through a group of bins ordered by shooting day, each “culled” timeline region would be in day order — whether that is the preferred order or not. Then you’d start moving clips around to get them in the preferred presentation order.

I have done that same procedure in Premiere but I don’t see how it’s any faster than marking range-based favorites with the skimmer in FCP X, then appending those to the TL in one step. In either case you must reorder the material once it’s in the TL. The issue is what’s the fastest possible method of blitzing through the raw footage and extracting the selects. On any hardware I’ve tested the skimmer responsiveness is much faster, especially with H264 4k material.

I agree the lack of a “play through” option in the FCP X Event Browser is a mild deficiency. If using only JKL within the Browser, you have to press down arrow to advance to the next clip. A sequential “play through” option is a commonly-requested enhancement. However — the optimal working Browser working mode is skim with the mouse, click/drag to mark ranges, and Favorite those with the left hand “F” key. If the Browser sort order is by date, they are appended to the TL in that order — just like your procedure, except in a single step.

Mouse-based Browser range selection may not always give the desired precision, leading to JKL which then requires manually advancing to the next clip. OTOH you don’t necessarily need that degree of precision since you aren’t making a cut but marking a range. It can always be adjusted in the timeline.

I don’t see how working sequentially through a group of bins, adding them to the timeline and deleting rejected regions produces more “linear”, usable results any quicker than marking favorites in the FCP X Browser and appending those to the TL in one step. For a “fast first assembly” task over a lot of material, the FCP X Event Browser seems faster.

[Joe Marler]

[Jeff Kirkland] “I’m on the verge of picking up a couple of XC10s for a project so I’m curious.”

I would suggest you also evaluate the Sony RX10 II. One of our camera operators got one for personal use and it looks pretty good. It also shoots internal 4k, has the same 1″ sensor size and a constant-aperture f/2.8 lens. I think FCP X supports its codec right now.

[Joe Marler]

Stacy, you have selected a good machine and will be happy with it.

Yes it’s best to use media on an external drive. This advice originally was based on performance and splitting the I/O workload. It was also based partially on the idea that editing required a low-compression codec, which resulted in high I/O loads.

Since Adobe’s Mercury Playback Engine and near similar performance by FCP X on native camera files, this changed the picture by enabling much editing to happen on high-compression native files. This reduces the I/O load but increases CPU load.

Your new iMac has over 1,000MB/sec I/O bandwidth to SSD, which also changes the “don’t overload the boot drive” advice. However 1TB is too small to fit much video on. So the practical steps are unchanged — put the media on external drives. That said you could put a small project on our system drive — it definitely has the bandwidth.

But — FCP X can generate a lot of space-consuming files: render, transcode, optical flow, etc. So just summing up your projected media files is not an accurate indication of your total space needs.

A good inexpensive app to help watch your space consumption is Final Cut Library Manager:
https://www.arcticwhiteness.com/finalcutlibrarymanager/

Although the 2015 iMac and FCP X are capable of doing three-camera multicam on native camera files, in some cases it might be a little more responsive if you transcoded to proxy or optimized media. This can happen on either import or afterward. Optimized media is about 8x the size so I/O load will be higher and space consumed larger. It is not necessary to do this and I have edited lots of H264 1080p multicam on a late-model iMac. It is a personal preference.

Re external drive, you can get by with almost anything, but a higher-end, faster drive is generally better for professional production use. You do not normally need SSD, esp if editing H264 camera native files. SSD is expensive, the storage is small and the I/O rate typically is not needed. Anyone can verify this by examining I/O statistics with Activity Monitor or iStat Menus when doing editing operations, including import, rendering and export. However if SSD fits within you space and size budget, there is nothing wrong with it — it’s very fast. Just keep in mind that both internal drive and any external drive (including SSD) should be backed up to another drive. So really you are talking about two drives — one for media and one for backup.

I would definitely not suggest a bus-powered USB portable hard drive. They are usually 5400 rpm and quite slow. However the 1TB HGST Touro S is pretty fast (for this class): https://www.touropro.com/en/product/touro-s/

One of the fastest USB bus-powered drives is the 4TB Seagate Backup Plus Fast. However it is internally two drives in RAID0, so you’d definitely want it backed up:
https://www.seagate.com/external-hard-drives/portable-hard-drives/performance/backup-plus-fast-hdd/

The ideal practice is to have two different forms of backup — say one constantly-connected hard drive for Time Machine and another drive you connect as needed using Carbon Copy or similar software. Yes, that would be three total drives — media, TM backup and CC backup.

Many editors use a Thunderbolt RAID array of some type. This isn’t mandatory and you might get by with a single fast hard drive for camera native material. However your storage should ideally be on the quality and performance scale of your machine and you have a very nice fast computer. Even though RAID5 can tolerate a single drive failure, it also should be backed up.

A lower-end quality drive array with good performance is the G-Tech G-Raid. It is available in 4, 6 and 8TB sizes. It is two drives in RAID0 so you’d need to ensure it’s backed up. You could back it up to another one or a similar capacity drive.
https://www.bhphotovideo.com/c/product/858463-REG/G_Technology_0G02272_8TB_G_RAID_External_Hard.html

Keep in mind starting with El Capitan, Apple removed RAID support from Disk Utility, so everyone using software RAID0 drives will have to either (a) Use terminal to format and verify them, or (b) purchase SoftRAID Lite:
https://softraid.com/pages/features/softraid_lite.html

The Promise Pegasus series are very popular:
https://www.promise.com/Products/Pegasus/Pegasus2/R4

OWC makes the Thunderbay 4 which is relatively inexpensive and has good performance for this category:
https://eshop.macsales.com/shop/Thunderbolt/External-Drive/OWC/ThunderBay-4

CalDigit drives and arrrays also popular: https://www.caldigit.com/products.asp

[Joe Marler]

[Zach Smith] “I’m importing from a P2 card that has footage shot on the Panasonic ag-HPX250 and the recording format is set to AVC-100/60i.”

I have imported a lot of 4k H264 stuff from the AG-DVX200, and don’t see that behavior. However I don’t import ranges from the card but copy the .mp4 files to HDD and import from that using “leave files in place”. The import is instantaneous with no major additional space used in the library. Range-based import only works with the original card structure.

I realize this may not be the optimal (or even proper) procedure and maybe isn’t helpful to your situation. However I just wanted to let you know we are using a similar camera and what our results were.

Since FCP X is so fast at skimming and ranking material, our overall approach is to just import everything with “leave files in place”. If it’s a rejected clip, we do that solely in the Event Browser, not by viewing and rejecting some outside and some inside the editor. This doesn’t take any more space since we have the files on local HDD anyway. We redundantly back up each day’s shooting in the field and save every clip whether an outtake or not.