[Joe Marler]

[Brian Thomas] “I’d like to make it appear as a juddering zoom-in”

MacBreak Studio did something called “Spooky Jump Cuts” using the retiming option “Jump Cut to Markers”. Something like this might work for a zoom:

https://www.youtube.com/watch?v=fFb8CGRXaUI

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]

You have a synchronized clip which is a container for two base clips. As part of changes for the new “roles” functionality, if you want to change an audio component for a sync clip, you must right-click on the clip in the timeline, select “Open Clip”, then in the Inspector change it to Dual Mono, then in the timeline click the timeline history “back arrow” to return to the parent timeline.

It is similar with multicam clips except you must open it in the Angle Editor. It is a little less discoverable with sync clips because unlike MC clips they won’t automatically open from double-clicking on them.

There is some more info here, but it doesn’t specifically explain this situation: https://support.apple.com/en-us/HT207124

[Joe Marler]

[Jenn Lindsay] “how to make the audio dual mono or stereo…..

There’s no “Channels” menu on the audio configuration sidebar. And the Angles editor is a little beyond me”

In 10.3x it’s the same as previous for single camera material. For multicam it changed — in that case you must open the multicam clip in the angle editor, click on the camera angle you want, then in the Inspector audio tab, click on the waveform and you’ll have the option for dual mono.

[Joe Marler]

Another solution is to sync it with PluralEyes, which does continuous correction for sync drift over the entire take: https://www.redgiant.com/products/pluraleyes/

[Joe Marler]

[Bill Davis] “I get the halo effect and all, but the movie biz is numerically small potatos when it comes to an NLEs development and worldwide adoption. “

Avid has a firm lock on Hollywood — maybe over 95%. However there is little money there. Avid’s annual net revenue from video products and solutions has declined each year since at least 2011, and is currently about $201 million, or 66% of what it was in 2011. As recently as last year there were news reports of continuing layoffs. Apple’s revenue from watch bands is probably more than that.

Total Hollywood box office revenue in 2015 was about $38 billion but Youtube is projected (by one analyst) to reach $27 billion by 2020:
https://www.tubefilter.com/2016/04/15/youtube-estimated-revenues-27-billion-2020/

It’s great to hear about either Premiere or FCPX used in feature films, and the products benefit from that. However if Apple had not “rescoped” FCPX for the broader market, I’m not sure it would exist today — just not enough revenue in the upper segment to justify the development effort. OTOH Apple (unlike Avid and Adobe) makes a lot from hardware sales and if FCPX is viewed as a Mac sales multiplier, the total revenue benefit is a lot more than the software.

[Joe Marler]

[Dominic Deacon] “It honestly takes about half an hour to an hour to put a pc together. …What’s unproven though? The cpu will come from Intel….The GPU will come from nVidia…These are all battle tested parts….If any part of my pc fails it’s under warranty….I built my most recent pc 18 months ago. Over that period it hasn’t crashed once- and it’s massively overclocked- let alone any hardware issues. No reason for there to be either. They are all quality parts….”

In a previous job I built hundreds of custom PCs for corporate use. On an *individual* basis a certain configuration might seem to work. But stress testing dozens of them under wide conditions would often reveal intermittent systems integration issues involving firmware, software, FGPA/ASIC versions, etc. This in turn required engineering-level interaction with the suppliers.

Companies like Lenovo or Dell are not just slapping together parts. The so-called “standardization” of modern PC components does not magically avoid all integration problems. There are engineers at those companies using expensive test instruments like this logic analyzer to debug those: https://photos.smugmug.com/photos/i-52kXRLG/0/X3/i-52kXRLG-X3.jpg

A single individual building his own PC does not have access to those resources. Yet standardization is good enough that you can often get away with building a PC and it usually works pretty well. I am typing this on a PC I built myself. However you can also get involved in bewildering, time-consuming debugging situations where the problem cause is unclear, esp. for intermittent and infrequent issues.

Re why doesn’t Apple just build an ATX-form machine, one reason is they are looking to the future. ATX is an old paradigm rooted in the 1990s, and if modern designers were approaching the task today without that historical baggage, they’d do something different. You can see some discussion of this by industry professionals on RealWorldTech.com: https://www.realworldtech.com/forum/?threadid=163257&curpostid=164090

Each Thunderbolt 3 bus supports bi-directional transfers at 40 gigabits/sec (80 gigbits/sec total), which is faster than 4-lane PCIe 3.0, and real-world tests have shown that transfer rates above that haven’t yielded much practical benefit (for GPUs): https://www.tested.com/tech/457440-theoretical-vs-actual-bandwidth-pci-express-and-thunderbolt/ That said the PCIe interface (as opposed to a row of PCIe card slots) is definitely useful for high-speed SSD.

So while ATX and a row of PCIe card slots are comfortable and familiar, they are rooted in the past and if designers could start afresh today they would use different technologies. That is what Apple’s doing.

[Joe Marler]

[Pippa Young] “using the same camera (Canon XF100)…After updating to the new FCPX (v 10.3.2) the next video I exported had those horrible interlaced lines on them which I’d never seen before. …that footage I’m sure is progressive is interlaced and coming out horribly. I tried using the de-interlace button on the footage, changing the timeline to progressive, keeping it as interlaced, exporting as both. Nothing works! What is really strange is that in the original file the footage looks fine, and in the FCP window it appears fine but when exported it looks so much worse…”

This is a tricky area. The XF100 can shoot either interlaced or progressive footage. Is there any possibility it was switched to interlaced for any of this footage, or has been interlaced all along? You can’t trust visual playback behavior because various playback tools act differently. E.g, VLC defaults to “don’t deinterlace”, whereas it appears QT Player 10 automatically deinterlaces — at least some of the time, likely based on header fields in the video file.

Even if you select 1080p/25, your XF100 may be using “Progressive Segmented Frame” (PSF) encoding, whereby 25p content is encoded as 50i. Editing and playback software may not always interpret this correctly: https://en.wikipedia.org/wiki/Progressive_segmented_frame

It’s often good to inspect original vs exported video file characteristics with a tool like Invisor: https://itunes.apple.com/us/app/invisor-media-file-inspector/id442947586?mt=12 or MediaInfo: https://itunes.apple.com/us/app/mediainfo/id510620098?mt=12 I prefer Invisor since it can display several file headers side-by-side in a spreadsheet-like format.

I did some brief tests with some older 1080i AVCHD content from a Canon HF-G10 and I see similar behavior to what you report — on both 10.2.3 *and* 10.3.2. In both cases dropping a 1080i/29.97 clip into an “automatic settings” project creates 1080i/29.97 settings. It looks OK within FCPX but has obvious interlacing comb artifacts (as seen by QT Player) if exported — no matter what deinterlace and field dominance settings are used. With VLC you can manually invoke deinterlacing and make it look better. In this case QT Player does not automatically deinterlace.

I tried all combinations of FCPX deinterlace and Field Dominance settings on both FCPX 10.2.3 and 10.3.2, and none of them prevented interlaced-looking output — when viewed in QT Player or VLC without turning on deinterlacing.

If I drop the 1080i clip into a 1080p project, then export with the default settings (deinterlacing off), it looks OK. Likewise if I re-wrap the 1080i AVCHD content with EditReady before importing it, the exported file looks OK in QT Player.

Importing the same 1080i clip into Premiere CC 2017, letting it automatically create a sequence then and exporting it to H264 looks OK.

There are two key metadata fields in the video file header, Scan Type and Scan Type – Store Method (thereafter shortened to “Store Method”).

Examining the video file headers with Invisor shows my original 1080i file is Scan Type = Interlaced, Store Method = “Separated fields”, and Scan Order = Top Field First. Note these fields are advisory — they should (but don’t always) indicate the physical encoding. Playback and editing apps can (but don’t always) read these fields and take proper action.

The file exported from Premiere is marked Scan Type = Interlaced, Store Method = Separated Fields, whereas the files exported from a 1080i project in FCPX 10.2.3 or 10.3.2 are marked Scan Type = MBAFF (Macro Block Adaptive Field Frame), and Store Method = Interleaved Fields. MBAFF is a special feature of AVC encoding which mixes progressive and interlaced content in a single frame. I don’t know if the data is really encoded that way or if FCPX simply marks that in the file header.

With the 1080i file imported to a FCPX progressive project, the output file is marked Scan Type = Progressive, Store Method is blank. The actual content is probably deinterlaced during render/export.

My first guess would be FCPX is erroneously marking the metadata field Scan Type as MBAFF, Store Method as “Interleaved Fields”, and isn’t physically deinterlacing the footage when you click the deinterlace check box. Hence the output files are still interlaced but the playback software reads these fields and is fooled into not deinterlacing for display purposes. When I upload a test clip to Youtube or Vimeo it looks OK so they must somehow be deinterlacing it regardless of the file header. It superficially looks like an FCPX bug, but it could be a “standards issue” regarding how software should interpret those fields, maybe in the case of Progressive Segmented Frame content.

My recommendations: Verify whether your original camera files are interlaced using a tool like Invisor. I am guessing they are. If so you can try putting those into a manually-created 1080p project, export them and see how they look.

I will study this a bit more and give Apple a call. Fortunately my group hasn’t shot any interlaced content in several years but we have a lot of archival interlaced content.

[Joe Marler]

[Jenn Lindsay] “Would it be a terrible idea to update my iMac version to 10.3 while I’ve got some active projects?
“

There is always some risk in upgrades, and 10.3.x is a major upgrade. Also the UI has some changes so it will take a day or so to become accustomed to that. I was forced to upgrade a 5 TB project to 10.3.x because of a plugin rendering bug in 10.2.x. The upgrade mostly went OK but I later hit a crash on import problem that took a while to figure out. Overall I like 10.3.x better.

If you are doing collaborative work with someone on 10.3.x, it may be worth upgrading — you’ll have to make that decision. Part of the upgrade process is evaluating what plugins you use and making sure those are updated and work properly in 10.3.x.

10.3.x has some nice features, also a few frustrating minor issues like the Browser layout reverting to the default each time FCPX is restarted. In general there are lots of useful improvements.

If you save a copy of your libraries and your current FCPX application, it is relatively straightforward to roll back, provided you don’t need to preserve work done on 10.3x.

[Joe Marler]

[andy patterson] “I will admit Premiere Pro 2017 did take a hit in real-time performance. Will it get fixed soon? Who knows? Premiere Pro 2017 is buggy because Adobe got rid of the beta testers. Adobe used to be a good company but not anymore. They are mediocre at best these days.”

Some say that Premiere CC has developed stability and performance issues, but I have also encountered plenty of these on FCPX. 10.3.1 introduced a significant performance regression (so-called “drag lag” problem) on longer timelines and even shorter timelines with many connected clips. It is supposedly fixed in 10.3.2 but it still seems a bit laggy to me. There are also an uncomfortable number of crash reports on 10.3.x, and I spent a week isolating one for Apple which involves an immediate reproducible crash on import.

Despite my previous post about FCPX being faster than Premiere on 4k H264 multicam, I just tested a 2 hr 2-camera 4k timeline on FCPX and it is unusably slow without proxy — on a top-spec 2015 iMac 27 with all media on a 4 x 2TB Thunderbolt 2 SSD RAID-0 array. Now that Adobe finally added proxy support to Premiere it is pretty fast on H264 4k. If you have to use proxy for 4k H264 on both Premiere and FCPX, you could argue what’s the actual performance difference?

Maintaining software quality in a complex product is like pushing a huge boulder up a hill. A moment’s inattention and it gets out of control.

Years ago of the CEO of FoxBase would walk among his developers with a stopwatch on a lanyard around his neck and demand impromptu timed performance tests of their code. I suppose that is not done anymore.

[Joe Marler]

[Tom Sefton] “I’ve also noticed vast performance differences from CC to FCPX with raw media on more recent cameras from Red.”

The performance differences are not limited to Redcode Raw. Editing 4k H264 multicam on my 2015 iMac 27, FCPX is more responsive at “Better Quality” than Premiere CC 2017 is at 1/8 resolution. When fast forwarding on a two-camera 4k H264 timeline, Premiere’s multicam monitor is more like a slide show — it updates about every five seconds. Likewise, JKL keyboard lag is much worse on Premiere. This is on the same hardware and using the same media files.

Of course you can compensate for these performance issues in Premiere by building a fast custom machine, but you can do the same thing for FCPX if needed. Here is an easy-to-build $998 machine that’s faster than a 6-core nMP with dual D700 GPUs: https://www.youtube.com/watch?v=I6ZJWPi_CBc

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.