Joe Marler
22
Apple Final Cut Pro X Coins
39
COW Coins
Forum Replies Created
-
Using PluralEyes 3.3.4, there was an XML sync problem which appeared around FCP X 10.1.2 or maybe 10.1.1. It is fixed in the latest version of PluralEyes 3.5.
-
[Nicolas Vallejo] “roughly 70% of it will be 60fps shot on a MKIII + 7D so 1280X720 is the highest output”
On the 7D, aliasing (stair stepping of diagonals) can be significant at 720. It is much less on the 5D3. The 7D does better at 1080 from that standpoint. Of course 60 fps gives smooth 1/2 speed slow motion, which is artistically useful for concert footage. OTOH FCP X retiming with optical reflow produces pretty good slow motion from 30 fps.
We shoot with a variety of DSLRs, including 5D3 and 60D (sensor/image processing similar to the 7D). Our general practice is shoot 1080p/30 and only use 720p/60 if we really need it.
This assumes that you as the editor can request the shooting frame rate. If it is already shot or the parameters determined by someone else, you’ll just have to live with it. You can easily mix 1080p/30 and 720p/60 material in the same sequence. The frame rate is automatically conformed, and they are both 16:9, so it’s a simple scale up or down to match one to the other.
Ideally you should shoot some 720p/60 test material with both 5D3 and 7D, especially looking for aliasing and see if it bothers you. If you have no problems with it, then 720p/60 might be OK. However as already mentioned 1080 gives more leeway to crop and recompose a bit in post.
-
The built-in audio sync feature of FCP X is mainly designed to sync a single audio and video clip, or two video clips. It works well for that. As you add more audio and video clips to a sync attempt, the reliability rapidly diminishes.
If you don’t know what audio and video clip goes together (say date/time wasn’t set correctly on the recorder), you can sometimes order them by length in the event viewer. That assumes each audio and video take were about the same length. That may get you close enough to briefly skim them and then sync each group of two.
If that isn’t possible, PluralEyes version 3 works very well. I have asked it to search hundreds of audio clips for a match to a single video clip, and it reliably finds it. I’ve also sync’d many audio and video clips in a single pass. Earlier versions were not as consistent, but the latest version works very well.
-
[Don Smith] Promise: “I suspect that PD1 is causing the problem also other drives detected error. So please back up the data immediately to avoid data loss.”
This is crazy. The RAID was bought on Dec. 12th. Promise should ship me a new one, let me save my data and ship back the defective unit. I don’t have 8TBs laying around to which I can copy the 8TB of media on the broken RAID. Rats.
Since I bought it from the Apple Store (online) I wonder if I can just return it? Hhhmmm…. Still, I would have the problem of saving all that data.”
The good news is this may explain your observed symptoms. The bad news is further isolation and replacement is required.
Promise sent me a complete replacement R4 but I had to twist their arm a bit. I originally ordered it from B&H. I was under the impression that items ordered from Apple are covered under their (usually more robust) support policies.
Whether now or the future, you pretty much need something to back up the R6 to. Just because it’s RAID 5 doesn’t mean it can’t fail — as you are now seeing. Also software problems, device driver bugs, etc. can damage data, and RAID 5 is vulnerable to those.
In my case I have an 8TB Thunderbolt G-RAID (RAID 0) which backs up my R4. Besides that I have separate Time Machine backups on a 6TB Thunderbolt WD MyBook Duo.
That said, even if you had another 8TB, it would possibly result in a double copy: From failing R6 to your backup, then transfer that to the replacement R6. If there’s any way you could talk them into shipping you a fully-equipped R6, you could do it in one step.
Sorry about your problems. The R4/R6 are great storage systems and after my initial problems I am very happy with mine.
-
[Don Smith] “And the Promise website is just awful….asked for a service report from the Promise utility and I can get that, but to upload it I have to go to my case on their website. Fine. Except no amount of clicking on the link will do anything.”
The Promise website is poor. It uses a pop-up which Safari blocks by default – without any message or advisory. The behavior is you click on your case number and nothing happens.
To prevent this go to Safari preferences->Security, and clear the box “Block pop-up windows”. Once open, the Promise support case allows you to attach files and system reports.
Re blue vs amber lights, on my R4 each drive has two lights. The top light is status and the bottom activity. I have never seen anything besides two blue lights per drive when it was functioning OK. If a drive fails, the top light turns red. When the failed drive is replaced, I think the top light turns orange/amber while the rebuild is in progress.
Your Black Magic performance results seem pretty good, but you should not have any amber or orange lights on any drive. The performance should generally be consistent. On my R4 I ran a burn-in test for several days using DiskTester, a professional test utility: https://diglloydtools.com/disktester.html
This following is unlikely related to your situation, but performance can be affected by RAID 5 stripe size. For test results of various stripe sizes vs RAID initialization time and performance tests, see this data: https://forums.macrumors.com/showpost.php?p=18576138&postcount=1
I’m not suggesting re-striping your disk array to improve performance, this is just FYI. However if in the process of debugging the situation you need to re-initialize the array, you might then consider optimal stripe size.
-
I just got the latest version of Plural Eyes. It is vastly improved over the version I used two years ago. I’ll test it more over the next few days.
-
[david mayer] “Does Plural Eyes require the
20 original small clips or can it work with the larger,
converted ones?”My recollection from using Plural Eyes on Windows Premiere Pro CS6: it syncs on a per-clip basis.
A significant limitation is it only does one sync point per clip — defined as the middle of the clip. If a long clip has sync drift it cannot compensate for that. E.g, one camera or recorder is running slower than the others due to poor master oscillator, or just a very long take as with lecture material. That would seem to be similar to your situation of concatenating multiple clips.
They formerly had a product called Dual Eyes which did sync drift compensation. I don’t know the status or availability of Dual Eyes after the acquisition by Red Giant Software.
Plural Eyes was finicky about audio levels, even when using its optional computationally-intensive “try harder” mode. E.g, if a DSLR with a scratch track from on-camera audio had poor s/n ratio, it just wouldn’t sync — even though you could easily sync the waveforms by eye.
In general my experience with FCP X audio sync is much better than Plural Eyes, but this is comparing to Plural Eyes as of two years ago.
As John said, you may need to cut each DSLR clip. If discrete clips, Plural Eyes can sync those with a continuous audio track.
-
[Jason Jenkins] “My Pegasus R6 (6TB) took around 24 hours to get through the setup, which was waaaay longer than the instructions indicated. No problem, just unexpected and disconcerting.”
On an 8TB Thunderbolt Pegasus 1 R4, I did extensive tests on this and reported my results to Promise. It appears the initial RAID 5 synchronization time is heavily influenced by stripe size. At the default of 128k, my 8TB R4 took over 83 hr (3.5 days) to sync. 256k and larger stripe sizes were much faster to sync. At 512k or larger, it took about 10 hr.
I tested three different R4 chassis and two different four-drive sets. I don’t know if the Pegasus 2 is any different. R4 chassis firmware: 5.02.0000.98. Drive sets were 4 x 2TB Toshiba DT01ACA2, with firmware of MX4OABB0. Using Promise utility, background sync rate was on “high”.
Details and data graph: https://forums.macrumors.com/showpost.php?p=18576138&postcount=1
-
[Poo Yan] “video I recorded with a GoPro Hero 3 at 1080p and 60fps…When I play back files in Premiere Pro, it’ll usually play smoothly for a few seconds and then drop to maybe 2 or 3 fps…I decided to throw copies of some of the video files onto an SSD and import to the timeline… the performance issues remained”
I frequently edit 1080p/60 video from a GoPro Hero3 Black in CS6.05 without *major* performance problems. It is slightly “laggy” when doing rapid JKL editing, but absolutely no problems at normal playback speed. Whether monitor resolution is 1/4 or 1/2 doesn’t make any difference. I do not transcode to Cineform or anything else. My system specs aren’t too different from yours:
Windows Home Premium 64-bit, i7-875K @ 4 Ghz, 16GB RAM, GTX-660 2GB, 1TB 10k rpm boot drive, 2x 2TB 7200 rpm Caviar Black data drives in RAID 0. Apps and scratch disk on 10k rpm drive, video files on 2-drive RAID 0.
If I turn off Mercury rendering it is a bit more sluggish, so make sure you have that on and it’s working. Add an accelerated effect like Fast Color Corrector, and make sure the render bar stays yellow. If not, Mercury is not working.
During the slightly “laggy” 8x fast forward and rewind states, perf. monitor shows it’s a multithreaded CPU-intensive operation with about 35 megabytes/sec inbound from the disk. There is plenty of additional disk and GPU headroom, at least on my PC. All 8 virtual cores are about 70% in this state. The limitation is on the CPU side, which could only be alleviated with more or faster cores, or an improved algorithm.
Summary of suggestions:
(1) As already suggested it’s always good to edit off higher speed drives, not variable-speed or 5400 rpm drives. Your SSD test seems to indicate that’s not an issue for this problem, but higher-speed drives are a good standard practice. You didn’t state the drive interface. Do not ever edit from a 5400 rpm USB 2.0 drive — they are just too slow.
(2) Update to 6.05. No known fixes for this, but you may as well
(3) Verify Mercury rendering is working and your GPU card name as indicated by gpusniffer.exe is listed properly in cuda_supported_cards.txt.
-
[Angelo Lorenzo] “You can then mix in the third track of audio independently of the two based on what you need it for.”
Could you please elaborate on the specifics? You mean do the two-camera multi-cam edit with “audio follows video”, then for each necessary audio edit, go back and manually edit both multi-cam audio and third audio track to switch in/out as needed?
Consider a two-camera lecture with frequent audience participation recorded on the third audio track. You’re saying each time the audience speaks you’d keyframe the multi-cam audio, lower it, keyframe the third audio, raise it, then revert at end of comment, then repeat each time?
Is there some easier way?
If multi-cam only allowed selecting the third audio track, it could all be done in a single pass.