[Joe Marler]

I have a top-spec 2015 and 2015 iMac 27 I use on large documentary projects with FCPX, mostly with 4k H264 material. The 2017 is very impressive — about 2x faster than the 2015 at ingest and creating proxies or optimized media, about 2x faster at encode to H264, much faster at scrubbing an H264 timeline. I don’t know why it’s so much faster — I assume the Kaby Lake Quick Sync is improved.

Overall it is not 2x faster at everything vs the 2015 — the GeekBench CPU and GPU numbers are a better reflection of the overall difference (which is generally modest). But for H264 it is hugely faster. If you have an all-ProRes workflow from acquisition to delivery, a 12-core D700 nMP still isn’t too bad. But for H264 on FCPX, the 2017 iMac is about 2x faster than the top-spec nMP on several key areas such as ingest, proxy creation and export to H264. With 4k ProRes the nMP is still about 2x faster on certain effects like sharpen and add noise than the 2017 iMac, so there are lots of differences based on what codec is used.

I have about 128TB of Thunderbolt arrays hooked to the iMac. FCPX 10.3.4 so far is handling 20 terabytes (about 300 hr) of documentary footage OK.

However there are cases where I still need more performance. Fast as FCPX is on the 2017 iMac, I still must generally transcode 4k H264 to proxy for multicam and the most fluid editing performance. If you have a high shooting ratio, that’s a lot of transcoding time.

During the early phases of editing, you don’t need the performance as much. But during the later phases, you end up rendering the project over and over, almost countless times. You are also adding very compute-intensive effects such as stabilization, Neat Video, Imagenomic Portraiture, Digital Anarchy Flicker Free, and hundreds of editing refinements — all those slow down the render and export. As delivery deadlines approach, it becomes critical how rapidly you can repeatedly render, export, and incorporate changes — on a timeline that is bogged down with lots of effects.

I had a 12-core nMP for a while and I really liked how quiet it was. But if your desktop is full of rotating Thunderbolt arrays, the computer noise is just one element among that. That said, when my iMac is transcoding for several days, I wish it was a bit quieter. It has been totally reliable but I wonder how running with all CPU cores pegged for days will affect long-term reliability and longevity.

A top-spec 2017 iMac 27 is pretty nice, esp. for H264. For 1080 material the 2017 iMac seems about all most people would need, at least on FCPX. 4k is another story, and especially H264. If you will ever edit HEVC material you’ll need hardware acceleration of some type which means Kaby Lake or later or whatever trick the iMac Pro uses (maybe AMD’s UVD and VCE hardware). The base level iMac Pro isn’t that much more and it will probably be considerably faster. How much so, nobody yet knows because the early reviewers (except Vincent Laforet) didn’t do any meaningful video performance testing, and that was on the 10-core Vega64 version.

[Joe Marler]

[brianlaz] “I’m in a timeline, and I Shift-F to match frame. I’ve identified a clip I’d like to use in another timeline. When I click on the other timeline the browser, poof–the clip is gone from the event viewer and I need to just search for it manually. This happens often enough that’s it’s a bit of a bummer. “

I don’t see it as “gone from the event viewer”: when I do SHFT+F and then switch to another project, the previous clip is still marked as a range with a thin yellow box around it. However it’s not positionally locked in place in the event browser.

If you want to mark a clip you find important, you can (1) Favorite it (2) Keyword it, or (3) Mark it with a marker and create a one-time Smart Collection to search for clips with markers.

Favoriting a clip is one key, keywording is one key combination if the keyword HUD is up (see MacBreak Studio 223, “Warp Speed Keywording”), and adding a marker is one key – “m”. For the marker case you’d have to create a one-time Smart Collection where text includes “marker”.

I agree it can be a learning experience to think of data in relational terms instead of navigationally. In relational terms there are attributes and queries for those attributes — the data has no intrinsic order outside of a result set. Humans tend to think of data as having an intrinsic order, e.g, “Brown” comes before “Smith”. But what if Brown lives in Texas and Smith lives in Alabama — maybe sorting by state is the most important. That’s why relational databases (like FCPX) have no intrinsic sort order. There are attributes, and queries for those attributes. Any order only exists in the scope of a query result.

It is common to scour past projects for material. That material wouldn’t be in the project (aka timeline) unless it had some good quality. It might be a useful enhancement if FCPX allowed selecting multiple clips in a timeline and doing “match ranges” on them all, then you could mark, favorite or keyword all those matched ranges in a single step. However it doesn’t support that so you can do it one clip at a time. This is still pretty fast when you use shortcut keys to favorite or keyword it. Unlike range selections those are persistent attributes that won’t evaporate if your mouse finger twitches.

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]

[Noah Kadner] “Ingest as is into FCPX- generate proxies- edit in proxy mode- switch to original quality mode for final edit output-done.”

In addition you can often import with “leave files in place” (depending on the media type). This makes the import much faster than copying to the library. Where tree-based media does not allow this, you can often rewrap with EditReady which is much faster than transcoding, then import the files in place.

By default proxies go into the library but there is a procedure for using external proxies, which FCPX generates outside the library. This is no faster but keeps the library file small so it can be interchanged among several editors. This procedure is described in Ripple Training’s 10.3 Media Management tutorial.

Editors have long sorted through material before ingest — because editing software was slow and clunky at browsing through material, plus there were few organizational tools inside. But FCPX is super fast at browsing content and has good organization tools. In many cases the fastest approach is just ingest with “leave files in place” and do most of the organizing inside FCPX. Disk space is relatively cheap — what’s expensive is time.

This procedure works well enough for me to edit a 20 terabyte 4k documentary on an 2017 iMac 27.

That said, the rate of transcoding from 4k H264 to ProRes proxies varies widely between machines. A 2014 MacBook Pro is OK but not that fast. A 2017 top-spec iMac 27 is *twice* as fast as a 12-core D700 Mac Pro, and about twice as fast as a 2015 iMac 27 at this task. But for smaller data sizes a 2014 MBP might be OK, especially if you let it run overnight.

[Joe Marler]

FCPX simply syncs at one matching point in the waveforms, it does not do continuous drift correction like Plural Eyes (which I think you have but had some problems with it).

This means the only thing maintaining sync over a long take is the inherent timebase stability in each camera. This implies your two C100s have a low drift rate, at least with respect to each other, and the Sound Devices field recorder (for whatever reason) does not.

The root source of the timebase is an inexpensive quartz oscillator, just like a cheap quartz wristwatch. Oscillators don’t have perfect frequency stability, which is why two quartz watches will gradually drift apart. Actual surveys show some quartz watches drift by 2 sec per day (much worse than the spec), so two watches with opposite drift could move apart at 4 sec per day. For two audio recorders this equates to 100 milliseconds per hour, easily enough to cause a lip sync issue on longer programs.

It is possible to make precision temperature-controlled crystal oscillators that have low drift, or use GPS-corrected timebases, but most cameras and audio recorders don’t use those. That said, I’ve seen inexpensive field recorders with a low drift rate. Maybe most of them are fairly stable but older or sick units develop drift.

I’d suggest testing your Sound Devices recorder individually vs each C100 or vs other cameras to see if it’s the “odd man out”.

Sound Devices makes good recorders but from a FCPX standpoint I actually prefer the Zoom F4 and F8 because they have iXML and produce a single polyphonic .wav file. This makes track identification easier in the editor. I don’t know why so few field recorders support iXML. This video shows the benefit:

https://www.youtube.com/watch?v=V7tcrNlnsQU

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]

[Evan Stanfield] “Will filming in 4k, importing in 4k and exporting from FCP at HD look better or the same as filming in HD, importing in HD, and exporting in HD…”

With every other factor held constant, and if viewed on a large enough screen, filming in 4k and exporting at 1080 can look better than filming in 1080 and exporting in 1080. I say 1080 because technically 720p/60 is also HD by ATSC standards, and because ABC, Fox and ESPN broadcast exclusively in 720p/60 HD.

However there are many variables and it’s easily possible to have better results shooting and exporting in 1080 vs 4k. If the 4k camera, lens, codec, lighting, cinematography or the distribution system to the final viewer is not good quality, then just using 1080 might be better. There are situations where 4k is just carrying around a lot of dead weight.

Another variation is if shooting 1080 on a camera optimized for 4k. In that case the 1080 image may be degraded vs. a camera designed specifically for 1080. E.g, my old Canon 5D Mark III shoots much better 1080 than my Sony A7RII, but the A7RII’s 4k down sized to 1080p looks better (even in very low light) than the 5D3’s native 1080p.

A good quality camera with a good lens that shoots 10-bit ProRes 4:4:4 at 1080p can look better than many consumer or even prosumer H264 8-bit 4k cameras.

However in many cases shooting 4k is a good idea, even if 8-bit 4:2:0 H264. If you don’t crop it and export to ProRes, it will supposedly become true 1080p 10-bit 4:4:4 on output. See white paper by Barry Green “The Benefits of Shooting in 4k”: ftp://ftp.panasonic.com/provideo/agdvx200pj/4kbenefits_techbrief.pdf

[Evan Stanfield] “…Is there some sort of standard, or loose rule of thumb around how much you can zoom in FCPX before it degrades the quality of the image? For example, could I take footage shot in 1080 could I zoom from a long shot to a mid shot without losing too much definition?”

If you assume the final viewer has 1080p resolution and the image is not degraded by encoding, UHD 4k can be zoomed 200% (using the FCPX scale factor) without losing effective resolution. If you are exporting 4k to 720 the limit is 300%.

If the original material is 1080p and the viewer is 1080p, you cannot zoom into this any without losing some quality. If the final viewer is using a smart phone via a streaming Youtube video, you might get away with a little zoom. In that case the viewer’s effective resolution might be 720p or less.

[Joe Marler]

FCPX might sync them if they are properly labeled. Each batch of clips from each camera must be labeled as a camera angle or camera name. Procedure:

1. In Event Browser at top left select List View (it’s the button above the red X at left)
2. In Event Browser, click on clip name column header to sort by name. All the clips from each camera or device should be grouped together. This assumes the clips from each camera have unique naming formats. If they don’t, you’ll need to tag them with a keyword in to enable batch selection. If the original clips still exist in separate folders on a hard disk, you can name the folders, then re-import the material with the import option “Assign keywords from folders”, which will automatically tag the clips from each camera.
3. In FCPX select a group of clips from one camera/recorder (Click on 1st, go to last and Shift-click).
4. In Inspector at upper right, click on Info button. If Inspector pane is not on, turn it on with Window>Show Inspector
5. At bottom of Inspector Pane, click the drop-down box and select “General”. This shows more details.
6. In the Inspector Pane, enter a camera name. When you press Enter it will add that camera name to all the clips you have selected in a single step.
7. Repeat steps 3 and 6 for each camera/recorder.
8. In Event Browser, select all clips from all the cameras you added camera name info to.
9. Right-click on the selected group and select “New Multicam Clip”, then set checkbox “use audio for synchronization”

After doing this open the multicam clip, disable audio from each lane except your main audio, turn on the angle viewer and check synchronization. If any clips did not sync properly you can re-try the sync within the multicam clip:

(1) Select the camera and audio icons besides the main audio lane. This makes it the monitoring angle
(2) Select the non-synced A/V clip
(3) At the drop-down menu to the left of that lane, select “sync selection to monitoring angle”. It will re-sync just the selected clip to the monitoring audio angle. You can also do with with multiple clips.

If this doesn’t work you are probably best off using Plural Eyes; it can sync almost anything:
https://www.redgiant.com/products/pluraleyes/?gclid=Cj0KCQiA0vnQBRDmARIsAEL0M1mr_bjCgXGXQN-NBcxc9h_RSo92tFChM_KBkzJ2BqgiSYOvjL755mYaAkRvEALw_wcB

[Joe Marler]

[Gabriele Sartori] “Quicksync is great but for now it seems to be a consumer technology….I would be happy if my 12 cores Mac Pro is as fast as handbrake when I use compressor. At least for now, my output looks better than what I get with Quicksynk….”

As you can see from these encoding tests on industry-standard material, there is virtually no visible quality difference between Quick Sync and multi-pass software encoding: https://joema.smugmug.com/Video-Tests/Quick-Sync-vs-Software-Encoding-Test/n-CksJjj/

The fastest way to encode H264 on a Mac is using FCPX on a 2017 i7 iMac 27, which is twice as fast as a 12-core D700 Mac Pro — I have tested that myself. Supposedly Compressor has been upgraded in recent versions to use Quick Sync and is allegedly as fast as FCPX but in my tests it’s never been as fast. However I haven’t tested it in some time.

If you want to see how 100% CPU utilization doesn’t translate into best performance, compare editing 4k H264 in Premiere CC vs FCPX. On Premiere all CPU cores are pegged at 100% just scrubbing the timeline, and FCPX is much lower. However Premiere is laggy and slow on the same Mac hardware (despite the higher CPU utilization) so the situation is more complex than just CPU.

That said, when encoding or decoding long GOP formats like H264, each GOP should generally be independent of all other GOPs, so should be a viable candidate for a dedicated CPU core. That said there are long GOP variants which have inter-GOP dependencies, and it doesn’t take much to kill multi-core scalability due to Amdahl’s Law: https://en.wikipedia.org/wiki/Amdahl%27s_law

Parallelism within a GOP is almost impossible due to the sequential nature of the algorithm. This was well described by Jason Garrett-Glaser, lead developer of X264, in this video lecture:
https://www.youtube.com/watch?v=uOOOTqqI18A

[Joe Marler]

[dallas kruse] “Sony a7s ii….shooting some performance shots tomorrow where a LOT of colored lighting is going to be used. Dancing lights. Fog. Effects. Etc….I’m wondering if I should shoot in 4k or stick to 1080?…I know the advantages of being able to zoom or crop with 4k footage but wondering if it’s going to make a HUGE difference in output quality?”

I would NOT shoot 1080p on a camera designed primarily for 4k, unless meticulous testing shows that camera’s 1080p output is good in all your conditions. This is totally separate from the issue of 1080p vs 4k on a camera designed for 1080p, like the 5D Mark III.

E.g, my 5D3 produces good quality 1080p, esp. in low light. My A7RII (in 4k Super35 crop mode) does at least one stop better in low light but both are very good. In that comparison the issues might be logistical handling of the data on a long program, don’t have to transcode 1080p whereas 4k requires transcoding to proxy for good editing performance, etc.

However the 1080p vs 4k decision can be totally different if the camera is designed mainly for 4k. A lot of such cameras don’t produce very good quality 1080p. To get a 4k sensor to produce this, the image must be pixel binned or resampled or line scanned or heavily cropped or something else. This often degrades the image quality dramatically, especially in low light or adverse conditions.

Another issue is 4k cameras may force you to a non-ideal codec when used in 1080p, such as AVCHD. IMO that alone warrants staying with 4k. In FCPX AVCHD can be a little tricky to handle, and I usually rewrap it with EditReady before importing.

I don’t remember the A7SII behavior on 1080p — you’ll have to research and test that. So the issue is not combining 1080p and 4k on the timeline — that is easy for FCPX. However in proxy mode 4k material is 1080p, but 1080p material is only 960 x 540, maybe not enough for some editing decisions. Since FCPX’s proxy mode is global — all files are displayed as proxy or none are — when using mixed 4k and 1080p material you might be forced to use optimized media, not proxy. This allows display of mixed 4k and 1080p while retaining good editing performance. However this increases disk space by about 6x over H264. IMO it’s easier to just use all 4k and transcode to proxy.

The main issue is whether a camera designed for 4k will produce a compromised image if set to 1080p mode, as compared with a pure 1080p camera. If there is the slightest chance of this it’s better to shoot in 4k. If your particular 4k camera produces excellent, non-compromised 1080p (inc’l low light, no aliasing, equal dynamic range, good codec and bit rate for that resolution, etc) freedom from aliasing, then the issue is storage space and transcoding to proxy.

[Joe Marler]

[Tim Wilde] “…what workflow people here use when faced with a ton of footage and they want to first cull out the unusable footage. I am not referring to simply using FCPX’s rating system, because my goal is to reduce the amount of hard drive space used so I can move the usable footage to an SSD for editing.

It’s occurred to me that I could assemble a sequence of the usable clips on the timeline and then use Worx4X to spit out the desired much smaller clips, but maybe there’s a better way.”

I’m editing a large documentary which is currently 12.5 terabytes and will probably peak at about 20 TB. Most of this is 4k H264.

The old way of handling this was scrub through the content before ingest, pick out the good parts and only import those. However the most valuable commodity in video production is time, and nothing is faster than FCPX at skimming and classifying material. So I just import everything using “leave files in place” which does not consume more disk space.

H264 4k material generally requires proxies for smoothest performance, so I create those. That takes a while but using a proxy-only workflow it’s a one-time cost by one person. The proxies can be passed to downstream editors, saving both time and disk space for everyone else. For details of a proxy-only workflow see Ripple Training’s 10.3 media management tutorial.

Re reducing drive space to allow editing on SSD, that normally is not necessary. In most cases on most computers, SSD does not contribute dramatically to real-world editing performance vs an equivalent spinning drive array, and trying to cram everything into an SSD eats the most valuable thing you have — time. I have an 8TB four-drive RAID-0 Thunderbolt 2 SSD array, and it’s very fast but in real world FCPX editing I can’t tell much difference between that and a spinning four-drive RAID-0 array.

If you are editing 1080p H264 you don’t usually need proxies or optimized media, and media is smaller than 4k or lower-compression codecs. If you are editing 4k H264 you generally need proxies which add about 60% to the space. Optimized media multiplies space requirements by about 6x, yet doesn’t improve quality any so I normally don’t use it. If you’re using ProRes acquisition, that’s nice because 4k can be edited smoothly without proxies but it’s typically too large to fit on most SSDs.

In general, time is more valuable than disk space, and SSD does not dramatically improve real-world editing performance for most situations. Therefore I would not usually recommend restructuring your entire workflow around cramming everything into an SSD.

You can usually import with “leave files in place” which is very fast and takes no more data than already used by the media itself. Exceptions are AVCHD and tree-oriented formats like XAVC-S. AVCHD should be rewrapped, preferably using EditReady. XAVC-S can be re-wrapped or the bare files moved to a folder and imported from there. This is not an ideal practice due to losing metadata but so far I haven’t seen a problem with it.

Re trimming clips to save space, for better or worse FCPX media management is currently designed around entire clips, not clip ranges. Clip ranges are presented as a database “view”, but the backing clip is the entire file. There is no simple way to trim these, although that’s a long-requested enhancement.

Yes you can copy a bunch of stuff to a timeline and export that or use Works4 X on a timeline (with several limitations): https://worx4.com/Worx4-X/worx4-x-features.html

However that is a lot of time and work to save what is a fairly cheap commodity — disk space. If your goal is driven by the perception that SSD is important for editing performance, this may not be correct. Your most time-effective solution might be just using a sufficiently fast hard drive or spinning drive array, accepting how FCPX is currently designed, and working with its strengths.

[Joe Marler]

[Maximilian Richards] “…I spotted all my media (800 clips) and selected ranges on my favorite clips. I also rejected some clips but also did leave some clips untouched. Now I want the ranged clips only added to my timeline…How do I add all (and only) ranged clips into the timeline? Obviously cmd+a also selects “unranged” clips as a whole, which I prefer not to add to the timeline.
Any of you guys can think of a trick so I don’t have to manually select all the ranges? “

Apparently what you’ve done is mark a bunch of ranges then unselected them, leaving a thin yellow box. Now you’re trying to select only the ranges you marked, but not the unmarked ranges. To my knowledge there’s no single-step method of doing this.

The design of ranges is similar to Finder where you select several files out of a larger group of files. That is intended as a short-term selection which you then immediately take action on. In Finder you might move or tag those files you selected. If you accidentally deselect those files there’s no method to select just those again. FCPX behaves similarly.

The intention of marking ranges in FCPX is you take immediate action on those such as tagging them favorite or keywording them. Keywords and ratings are persistent.

You can re-select multiple unselected ranges by CMD+click or you can draw a bounding box around multiple clips by click and drag. This re-selects just those ranges you draw the box around. You can select discontinuous groups by SHIFT+ click and drag. If you accidentally select a full clip you can de-select that via CMD+click. This is similar to how Finder works in icon view.

Immediately after re-selecting the ranges you marked you should tag those favorite or keyword them.

After tagging a group of ranges “favorite”, you can save those as a keyword collection: show only favorites, CMD+A to select, then keyword those, then U to unfavorite them. This frees up the favorite tool for new use yet preserves all those you marked favorite.