[Joe Marler]

[Ken Bennett] “How are your organizing to media for a weeding? Library, Event and project. Would like to know you MO for wedding.”

There is no universal answer for this, and I’m not sure there’s a consensus on what organizational method is best. Projects are just FCPX terminology for timeline, so use those as appropriate. Depending on the deliverables, you might have a project (or timeline) for a concise web-based summary, a longer one for the bride & groom, maybe even specific projects for rehearsal and reception.

Hypothetically you could import content to a separate event for each discrete activity related to a wedding. E.g, separate events for bridal shots before the wedding, for the wedding itself, and for the reception.

But even that is not really necessary now that FCPX has library-wide smart collections and searching. Assuming all the content is going to be rated and/or keyworded, you could put all content in a single event.

You could probably do a small wedding using only range-based ratings — favorites, rejects and unrated — which gives three separate filterable categories without even using keywords.

For a larger multicam wedding you’d probably want to develop a keyword system which suits your taste and editing style. In general I’d suggest avoiding the tendency to be overly specific or use too many categories. Before assigning a keyword, ask yourself is it really needed. E.g, at first you might tend to keyword wedding vs reception but if the camera clocks are all set properly the reception will naturally be sorted chronologically after the wedding.

This is common sense, but on any multicam shoot always ensure that each camera’s time of day clock is properly set, also the clocks on any audio recorders. Doing that one step will solve many problems in post production.

This MacBreak Studio on “warp speed keywording” is good:

https://www.youtube.com/watch?v=azJ4J41JaZk

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]

[Grant Peacock] “choking every time I launch a library that contains just over 200GB in total material. I had split the library out in 2 approximate halves when I suspected that the size of this library might be the issue, but the issue persists each time I open again in FCP X…CPU activity is basically at idle…editing is impossible and the cursor lags while typing.”

By “choking”, what is the exact behavior? Is it slow to open the library or slow to scroll down through the Event Browser in filmstrip mode, or won’t scrub through the timeline quickly or the skimmer itself is halting and slow, or what?

Create a new project within that library and add a few small clips to it. Is that slow? If not then maybe some timeline effect in your main project is slowing things down. However this would normally manifest as high CPU, which you’ve already ruled out.

If you have background rendering enabled, try disabling that.

During these slowdowns, check your I/O using Activity Monitor. If bytes/sec or IO per sec are high, this could imply something is doing lots of background I/O. If they are both low and CPU is also low that’s not a good sign. It could imply some interrupt or hardware problem. I would suggest unplugging all external devices and running Apple Hardware Test: https://support.apple.com/en-us/HT201257

Also try running Disk Utility repair option on your main drive.

If no errors, try to boot FCPX but keep all external devices disconnected — USB hubs, Firewire devices, external hard drives, etc. Evaluate the performance like that and see if it changes. If it does, the slowdown might be caused by a malfunctioning peripheral device flooding the system with interrupts.

Yet another item I’ve sometimes seen is runaway or malfunctioning Spotlight index. I suggest try turning this off and/or rebuilding spotlight indexes: https://support.apple.com/en-us/HT201716

If you have any backup task running such as Time Machine, try temporarily disabling that.

My largest library is 6,718 clips comprising 138 hr of material, about half H264 1080p and half UHD 4k, 3.21 terabytes. That is on a 2015 top-spec Mac Pro with the media on a 16TB OWC Thunderbay 4. All content was imported with “leave files in place” with no proxy or optimized media generated. Performance is normally OK but sometimes the Event Browser can get bogged down generating the thumbnails.

After doing the selects and proceeding to the editing phase I often (but not always) create proxies for 4k. I could never create proxies (much less optimized media) for all the content — that would take forever and would be too big.

If your SSD is 70% full you likely don’t have much space to create proxy files for everything, since they are (very roughly) 60%-80% the size of the original media files due to the lower compression.

[Joe Marler]

[Scott Thomas] “One idea is that Apple will go with an AMD SOC (System on a Chip)
https://daringfireball.net/linked/2016/08/26/amd-soc

Apple is developing their own ARM based CPUs at a breakneck pace. What if they got Intel to fabricate them in their foundries?
https://daringfireball.net/linked/2016/08/26/intel-arm….Maybe we are going to see Mac move to ARM, if the history is any guidepost?”

There has been informed commentary about ARM-based Macs. However there’s also been lots of wild speculation based on misunderstood reports, like iPhone 7 Geekbench performance is faster than one core of a 12-core nMP. In fact the multicore performance of my iPhone 7 is slower than my 7-year old Windows PC with a six-generation-old four-core i7-875K. My 2015 iMac 27 with a 4Ghz i7-6700K has 3x the multicore Geekbench performance of my iPhone 7. It is amazing the iPhone 7 is that fast but it is still a lot slower on real world multithreaded workloads than a seven-year-old PC.

When you are starting from lower performance (like ARM), it is easy to make big gains. Intel experienced that decades ago. Those gains cannot continue whether ARM or x86. You encounter fundamental architectural issues with superscalar instruction decoding, leakage current and fabrication. ARM (including the Apple-specific ARM designs) is making more rapid progress because their performance is lower so they haven’t yet reached that point of diminishing performance. When they do their performance gains will probably taper off, just like Intel.

If you were willing to give up Windows compatibility, an ARM-type CPU could (performance wise) replace the x86 in lower-end Macs. I’m sure Apple is evaluating that. It would fragment the product line, creating x86 and ARM-based Macs. It would greatly reduce the software catalog for ARM-based Macs. They would essentially be like the now-defunct Windows RT — a special version of macOS for ARM for which there are few apps. Apple might judge the tradeoffs as worthwhile but I doubt it.

But this thread is about the Mac Pro, and I don’t see any conceivable way ARM could compete with Intel up in the Xeon performance space.

An alternate approach is maintain x86 compatibility by partnering with AMD and Apple taking over their own line of x86 CPUs, similar to how they did with ARM. That is slightly more conceivable, but the gap from AMD’s projected performance to Intel’s Xeon performance is huge.

Based on Intel announcements, rumors and leaks, we expect significant Xeon upgrades in the relatively near future, including possibly on-chip programmable FPGAs. These could theoretically be reprogrammed by developers to execute application specific logic in hardware, e.g, maybe video encoding. It is conceivable the Mac Pro delays are to sync up Intel’s upcoming products with an updated Mac Pro design, plus of course the new 14/16nm GPU technology which only recently became available. Then there is Thunderbolt 3, USB-C, updated 4k monitors, etc. The production-quantity availability of all those items must converge within a specific timeframe to make possible a new workstation.

[Joe Marler]

[Jenn Lindsay] “I synced the external audio manually (see screenshots below). But the footage with audio now exists in a timeline instead of in an event, so I can’t append keywords….What is the best way to get timeline footage condensed or exported/reimported so it can be intact in an event and eventually be keyworded?
“

This is some more info on what Jeff and Noah said. Just because you are using a clip in a timeline doesn’t mean it cannot be keyworded. Make sure you have View>Show Used Media Ranges enabled, then do CMD+1 (Select Event Browser), then CMD+F (Open filter pane), and press the + key to filter on used media ranges. This will show in the browser only those clip ranges you have added to the timeline. You can also create a Smart Collection for the used clips, that way it’s handy and always ready. Those used clips can be keyworded however you want, even though they are already used in a timeline.

Remember to cancel the filter when you are finished (note the orange checkbox at the upper right of the browser pane).

However a complication is since you manually synchronized the clips in the timeline (probably using detatched audio), when filtering on Used Media this will show the separate audio and video clips in the browser. While you can keyword ranges within those clips, they are not synchronized so of little use.

As Jeff said it’s probably best to create a compound clip from your current timeline, that puts a single synchronized clip in the Event Browser, where you can keyword ranges and favorites. Those can be queried and used in subsequent development of the project, even though they are already used in the timeline.

Given this situation in the future it might simplify things if you’d manually synchronize the clips in the Event Browser vs the timeline. In the browser, just select the clips, right-click and select New Multicam or Synchronized clip but de-select “Use audio for synchronization” (since your video clips don’t have any audio). Then double-click on the non-synchronized clip in the Event Browser and move the clips to manually sync them. They will stay that way. The resultant manually synchronized clip can be range marked, keyworded, etc — all before adding anything to the timeline.

[Joe Marler]

[Mandy Leonardo] “I’m using Shift and selecting. Option select doesn’t do anything.”

In Macs, CMD+select is for selecting separate items. It is similar to how CTRL+select works in Windows.

By “When I select the part of the audio I desire”, are you trying to select a limited range on a longer audio clip and sync the video to that range as a discrete synced clip (ie file)? If so that is not how sync works, whether in FCPX, Premiere or Plural Eyes. They all create a single synced “clip” regardless of how many source clips.

If you simply want to sync multiple video clips to a longer continuous audio clip, that is a common procedure. Frequently an audio recorder runs the entire time and cameras are stopped and started. In FCPX the procedure would be select them all then create a single multicam or synchronized clip, using audio to sync. Then if you want “separate synced clips”, just select ranges from that big sync/multicam clip and favorite them or keyword them. Those ranges can searched, added to a timeline and handled as if they were discrete clip files or separate synced clips.

There may be long gaps between the video clips, say if the audio recorder was left running over a lunch break. However the skimmer and browser in FCPX are so fast this is normally no problem. You just zoom in and mark the ranges in the Event Browser, then favorite or keyword them. It may help to turn on audio waveforms in the browser by View>Show Waveforms, also stretch out the filmstrip view by using the slider at the bottom right of the browser.

If you don’t like the long gaps I suppose you could open the synchronized clip in the timeline, cut out the long “gap” clips, then make similar cuts to the audio clip, but that seems unwieldy and not how FCPX is designed to work.

Likewise you could make a big synced clip with multiple video and one long audio file, add the whole thing to the timeline, blade each shorter synced sub-clip, then add each back to the browser via creating a compound clip of each one. However this also would be unwieldy and not how FCPX is designed to work.

I am speculating on a lot of the above, just trying to guess what you may have in mind. In general the best procedure in FCPX is do most of your initial work in the Event Browser and make extensive use of its ability to mark and tag keyword ranges.

[Joe Marler]

That is a very good machine — I used one for two years until I got my 2015 iMac 27. It doesn’t have the Retina screen which ironically makes more difference on text than graphics. Before I redeployed it to another editor, I did a bunch of tests on it vs the 2015. It is not that much slower: https://photos.smugmug.com/photos/i-3mL6T5r/0/O/i-3mL6T5r.jpg

IMO for comfortable H264 4k editing you ideally need faster machine than a top-spec 2015 iMac 27, even considering the performance advantages of FCPX. However I edit 4k every day on mine and it’s usable. For this reason I don’t see the benefit of “stretching” your purchase and getting a 2015 machine. If you need a machine now it’s better to wait and see what the new machines look like since they will likely have a substantially upgraded GPU.

Whether 2013 or 2015 iMac you have to transcode 4k to proxy much of the time for best performance. So in that sense there is even less difference between the 2013 and 2015. I love the Retina screen but the non-Retina 2013 was a good machine.

You will definitely need a good disk subsystem, and something else to back it all up with. The OWC Thunderbay 4 is a good product and I use one along with various other RAID arrays: https://eshop.macsales.com/shop/Thunderbolt/External-Drive/OWC/ThunderBay-4

[Joe Marler]

[Stephan Scrofani] “Is there any way, other than manually key framing one person to an appropriate volume”

Here is a MacBreak Studio on using the compressor that’s built into FCPX: https://www.youtube.com/watch?v=OXxD0L9d7OU

Everyone probably knows this but you can also drag the range select (R key = range select mode) horizontally over the clip then drag that audio range up and down and it will adjust that region, putting a head/tail transition on the adjusted range.

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]

[Robin S. Kurz] “[Joe Marler] “The FCPX deficiency in how stabilization, optical flow and tracking plugins work is a dark, dirty corner of the product.”
? ? ?
What does that mean? Care to elaborate?

You’ll note there is no stabilization option for multicam, just like there is no optical flow smoothing, no Audition option, nor will a tracking plugin like CoreMelt’s SliceX work on multicam. In fact SliceX throws an error saying “cannot be used with multicam clips. Please switch to the angle editor and process the angle videos separately”.

What they really mean is you must backtrack and find the exact range in the base clip that corresponds to the edited MC clip, then blade that range, apply the effect then jump back to the MC clip and examine it.

The same thing applies with stabilization and optical flow smoothing. In general the procedure is:

1. Select clip on timeline
2. Shift+F to locate clip in event browser
3. CMD+1 to select event browser
4. Note clip begin/end time code and write them down
5. Double-click on clip in event browser to open in angle editor
6. In angle editor press CTRL+P and type in the timecode you wrote down
7. Blade that section of the clip, the apply stabilization
8. Return to main timeline
9. If stabilization needs adjusting (parameters aren’t available in Inspector for MC clip), then do it all again.
10. Repeat entire process for every MC clip requiring stabilization, optical flow, or tracking

In Premiere, the procedure is:

(1) Apply stabilization to edited MC clip — that’s it

Apple is very fortunate that Adobe has not done an “I’m a Mac, I’m a PC” ad for things like this. In an area where Apple claims ease of use, this is utterly embarrassing.

[Joe Marler]

[Bill Davis] “Wouldn’t we have known during prepping our clips for Multicam editing that we should apply stabilization (or have the AEs do it!) via Open in Timeline) before our multiclips are created?

Part of the X zeitgeist for me was to learn to engage my editors brain prior to the storyline stage to think about candidate processes that I can globally apply to clips early – so I get the benefit of more “perfected” resources in all my subsequent uses. “

That procedure can be a good idea for initial color correction. It is a lot easier to apply once to the base clip than to try and match dozens of little cut up clips on the timeline. However even in that case the final edited clips often need tweaking from a color and exposure standpoint.

Unfortunately doing stabilization or optical flow smoothing doesn’t work well on the base clips because it’s just too compute and time-intensive to run, and the final edited clip affects the fine-tuning parameters chosen.

The FCPX deficiency in how stabilization, optical flow and tracking plugins work is a dark, dirty corner of the product. Hopefully that will be fixed in the next version. Newbies to FCPX learn the product quickly at a superficial level because they’re not encountering things like that. They are likely not even doing split edits the “FCPX way”. They are mostly only tackling easy things. The magnetic timeline helps them, and the concept of the skimmer, event browser and even range-based ratings are logical and intuitive — especially to someone who hasn’t used a track-oriented editor.

Once the newbies progress to more advanced things, it becomes very hard indeed, and in some cases like these multicam behaviors, it is much harder than Premiere.

[Joe Marler]

[Scott Thomas] “The author was saying that Intel should be worried, not that they are in imminent danger.”

Actually I don’t think the author knew what he wanted to say. He metaphorically described the ARM threat to Intel as “…systematic downfall is all but assured…As The Matrix’s Agent Smith put it to an overweening police lieutenant, ‘your men are already dead.’ We may be experiencing such a moment in the tech industry today”.

Then he turned around and said “I’m not saying that Intel will be crippled or surpassed anytime soon”.

So which is it? Is it “all but assured” or not? He is not the first to notice ARM’s potential to compete with Intel in higher-end applications. This has been extensively discussed in the industry. Although the author concentrated exclusively on client side computers, in general the risk is greater on the server side, simply because what OS or CPU a server uses is mostly invisible to the client:

https://www.pcworld.com/article/3029740/components-processors/google-is-said-to-endorse-arm-server-chips-but-dont-get-excited-yet.html
https://venturebeat.com/2016/01/14/amd-finally-ready-to-crack-the-data-center-with-arm-chips/

A client side scenario with extensive legacy software is much more difficult. For one thing it would mean the end using Boot Camp, Parallels, VMWare, etc. In that case the choice would be between changing the CPU for all MacOS devices and writing off Windows compatibility, or splintering the product line between ARM-based and x86-based computers.

The rapid performance progress of ARM-based CPUs like the A10 can be misleading. It’s easy to make initial progress when you’re starting from a low point. The author of The Verge article did not indicate any awareness this conflict has already been played out before many times in the computer industry. For decades there have been repeated attempts to advance new CPUs with special instruction sets as vastly superior. In the 1970s it was high-level instruction sets which culminated in “high level language” machines which could directly execute in hardware a computer language. IBM’s Future Systems Project, the Burroughs B1700 and the Data General Fountainhead Processor (described in Soul of a New Machine) were examples of this. Then it went to the other extreme with RISC. Along the way there were attempts to use VLIW (Very Long Instruction Word) techniques to bypass some performance limitations of superscalar designs. In general the success of those approaches was limited relative to improvements in fabrication technology and to microarchitectural improvements for existing instruction set architectures.

There is no simple solution to this and I don’t see it as “all but assured”.