I have a 2TB Samsung 980 Pro PCIe NVMe SSD inside of an Acasis enclosure hooked via Thunderbolt to my Mac Studio M1 Ultra. The drive is formatted to APFS. It’s a fast drive, to say the least. However if I have a self-contained Library on the drive, sometimes the video will play back with the frames not in the proper order. Doesn’t matter what the codec of the source video is. If I export the finished project (also known as “sharing” or “rendering” in some circles) the resulting MOV file will have this improper frame order built in. I can actually step forward frame by frame in Quicktime and see the frames bouncing back and forth. The frames will sometimes play in the order of 1-2-3-5-4-6-7, just to give you an idea of what it’s doing. It doesn’t seem to matter if I export to h.264 or even to ProRes. However if I step forward frame by frame in Final Cut Pro, the frames are in perfect order as they should be. Most of the time it’s fine and exports fine and there’s no common denominator of the affected projects as they don’t differ from the ones that are fine.
Here’s the catch… If I copy the Library over to my USB 3 Sandsisk SSD, open it up and export from that with the same exact Compressor settings, it’s 100% perfectly fine. The Acasis enclosure is completely cool so it’s not overheating or anything. This seems to affect Final Cut Pro 10.6.2 and 10.6.3 as well as OS 12.3 and OS 12.4. Any ideas?
If it works fine with the Sandisk SSD but not the NVMe SSD in the Acasis enclosure, it seem pretty clear that the NVMe SSD or the enclosure chipset or the cable is bad. Or, possibly a formatting or driver issue, though not as likely. Right?
Well initially that was my first thought. But then a couple of weeks back I saw someone else post here about the same thing... jittery playback with frames in non-sequential order on the latest (at the time) version of FCP. So I figured maybe it's a FCP issue somehow. I said to myself "If this pops up again I'll make my own post about it". The Samsung SSD is my main editing drive now. Maybe it would pop up on other drives if I used them more.
Format is APFS. Like most external drives, it doesn't require a driver. I did try refomatting though.
Have you tried swapping with another Thunderbolt 3/4 cable? As I mentioned, the enclosure's chipset may not be working properly. Are there any firmware updates available for the Samsung SSD or the chipset in the enclosure?
It does seem pretty clear that it's the Samsung-Acasis enclosure combo that's causing the issue. If you can, try to work off of the Sandisk or get another external SSD. Then, use it long enough to see if the issue recurs.
I know that the Mac Studio is one of only three current Macs with true TB4 ports, with the Ultra having two TB4 in front and four TB4 in back, if memory serves.
You could try putting the Samsung SSD into another enclosure made by a different company.
I am ingoring the possibility (for now) that there is some sort issue with macOS or the Mac Studio HW...
Another troubleshooting tip... iStat Menus should be able to show the temperature history of your Samsung SSD. I have some Samsung SATA SSDs in a TB3 enclosure and iStat Menus "sees" the SSDs and reports their temperatures. It might be useful to monitor the SSD temperature and check it when the issue occurs. The Acasis enclosure may not get "hot" to the touch, but the SSD may still be getting hot...
I'm using the cable that came with the enclosure. Good call, I'll try another cable. I'll see if any firmware updates exist for either the drive or the enclosure. Not sure how to update the firmware if it's Windows only but I'll look.
iStat menu doesn't show me anything regarding the temperature of the drive, then again I have no clue how to use this software. But the enclosure is very cool to the touch.
There are no firmwares for the Acasis enclosure. There is one for the Samsung 980 Pro, but it's an ISO file and there's no way to apply it using a Mac. So basically no way to update the firmware on any of this hardware.
I looked on Samsung.com and found a list of firmware updates. Some of the older ones have a tag in the name that says "Mac". The one for your SSD doesn't seem to have a Mac version, though it may not matter.
For Intel Macs (may not work with M1 Macs): I know that for a lot of SSDs you can use a USB thumb drive or a DVD/CD disc, on which you load the ISO disk image (you might be able to use Disk Utility or Terminal commands).
With the CD/DVD loaded or the USB disk plugged-in, you then boot your Mac with the option key pressed and then choose "EFI Boot" (usually what it's called) when presented with the possible startup disks. You'd follow the instructions and the computer will restart/shutdown when done. You should have a backup of the SSD in case something goes wrong. It's definitely not a process for the faint of heart.
I don't know if the firmware update you downloaded will work with a Mac since it doesn't specifically mention that. You could ask Samsung support about this.
You could always take the SSD to a shop and have someone apply the firmware update for you using a PC.
I avoided going into more specifics, as the process is tricky.
UPDATE: Problem has manifested itself on the other USB SSD from SanDisk since I've been working off of that. I have used this disk for a couple of years with my old Mac Pro. Maybe it is a Mac Studio thing? Or a thing with the newer OS? On my Mac Pro I never went beyond Mojave and FCP 10.4.3 I believe it was.
I would agree that it could be a Mac Studio or a "macOS on Apple Silicon" issue (or FCP?). It does seem a bit less likely that it has anything to do with the external SSDs, unless there is a chipset issue, either the same chipset used or similar functionality for both external SSDs (and their controller HW) that is an issue. It reminds me of a _long_ time ago when certain chipsets in FireWire enclosures weren't ideally compatible with Macs/Mac OS (X). Unless it's an FCP issue, you're now in the realm of OS level drivers, HW, etc.
It seems like some sort of caching issue, where (as you said) frame-by-frame playback works fine, but after a bit of time (of constant playback or transcoding) the issue manifests itself. I know that some SSDs run really fast (near the marketing specs.) but can slow down after an initial burst of speed (same with hard drives) over time when doing a lot of I/O, due as much to the I/O protocol as any HW factor.
Unless there are some HW experts floating around here who could offer some advice, it might be time to contact Apple. Sorry I can't help further...
....I have a self-contained Library on the drive, sometimes the video will play back with the frames not in the proper order. Doesn’t matter what the codec of the source video is. If I export the finished project...the resulting MOV file will have this improper frame order built in. I can actually step forward frame by frame in Quicktime and see the frames bouncing back and forth. The frames will sometimes play in the order of 1-2-3-5-4-6-7....It doesn’t seem to matter if I export to h.264 or even to ProRes. However if I step forward frame by frame in Final Cut Pro, the frames are in perfect order as they should be...
Here are some troubleshooting suggestions:
1- Duplicate project to provide a backup.
2- When problem manifests during Quicktime playback, backtrack in FCP and see what clip that came from. Compare the clip attributes to the project attributes. IOW if that clip does not match the project, FCP will have to do rate conforming which can in some cases cause repeated frames (but not out-of-order frames). Admittedly that should also appear in FCP not just Quicktime, but it's an easy thing to check.
2a- Use Reveal in Finder to locate the clip file. Examine metadata to verify shutter speed. That can be often done with Invisor, but for some codecs it may require a vendor-specific utility. E.g, for Sony media you can use the free Catalyst Browse Utility. Purpose: certain shutter angles can produce an apparent duplicated frame. It should not be out of order but it's a standard thing to check.
3 - Change FCP preferences to show frame number: Preferences>General>Time Display>Frames. Purpose: have a consistent easy-to-read frame indicator for the below steps.
4- Add burn-in timecode to the project for both source and project timecode (2 Fx per clip). There is a built-in timecode effect. Project TC is based on timeline and source timecode is per clip. In both timecode Fx, select Format: frames. This will display frames not hh:mm:ss:frame. Apply both Fx to first clip, offset the two TC readouts vertically so they are both visible, do CTRL+C to copy Fx from that clip, then select all other clips and do Edit>Paste Attributes and paste only the timecode effects. Purpose: provide clip-level and timeline-level TC readout for subsequent troubleshooting, enables easy comparison to FCP and Quicktime frame readout (see below).
5- When doing any subsequent tests, use ProRes source files and export as ProRes 422 and view with Quicktime Player. Purposes: (a) All-Intra is a better troubleshooting format, (b) Avoids any possible problem from H264/HEVC accelerator (c) Enables frame readout in Quicktime Player (following step).
6- In Quicktime Player, click the timecode counter, which will switch it to frames (only works for ProRes). Purpose: inspect if the frame numbers shown by Quicktime also show incorrect order and whether they match the burn-in timecode frame numbers and FCP frame numbers.
7- Once a problem clip is produced with burn-in timecode, play in VLC, got to that region and examine frame by frame. VLC does not have JKL but press E repeatedly and it will go forward frame by frame. VLC does not have frame display but you have burn-in timecode for that. To skip backward/forward in VLC do OPT+CMD+left arrow or right arrow, then press E for single step forward. Purpose: since problem apparently only happens in Quicktime Player, test with VLC as alternate player.
8- Duplicate the project, open the duplicate, select all clips with CTRL+A then delete all Fx on all clips with Edit>Remove effects. Then add the above burn-in timecode effects to all clips. Export that project and see if the problem happens. Purpose: isolate if it's caused or enabled by certain effects. If it goes away when all Fx are removed, repeat above and successively remove Fx from regions of the timeline until you narrow down what Fx are involved.
9- Play problem clip with burn-in timecode on a non-Apple Silicon machine. Purpose: avoid possibility of a bug in ProRes accelerator that somehow manifests during Quicktime playback.
...The 30fps clips that the jittery-ness appears on are ProRes 422HQ. The files are recorded on an Atomos so there's no metadata.
There's a known issue with the Ninja V causing duplicated frames, at least from the Sony A7SIII, I don't know about other cameras. We saw this on multiple cameras and Ninja V recorders. In the case I investigated it was written to the source material and could be seen in both FCP and Quicktime Player. It was apparently duplicate frames but I can't remember the specifics. It seemed unique to 60 fps recording but maybe it happened at 30 and we never saw it, or had a different manifestation. It was intermittent and difficult to reproduce. It was not caused by a specific cable or SSD; we tested many of those.
If the camera shutter speed is not set to 180 deg, IOW if it's not 1/60 for 30 fps, that may create a similar effect.
You are right from a ProRes file you cannot tell the camera shutter speed. If the operator did dual internal/external recording, those internal files would be useful to troubleshoot this since they contain more metadata.
I vaguely recollect there was supposedly a workaround for new recordings (would not fix previous recordings). It was something like change camera frame rate to 59.94, then power up Atomos or something like that. I'll see if I can dig that up.
If it's not an Atmos thing but just 30 fps in a 60 fps timeline, that will hold every source frame for two 60 fps frame intervals.
If you don't need the same playback speed you could select the 30 fps clip and pick Retime>Automatic Speed. If you do need the same playback rate you could try selecting one of the 30 fps clips, go to Inspector and pick Rate Conform>Optical Flow, render that clip with CTRL+R, play it in FCP and see if it's smoother. If so export that clip and see how it plays in Quicktime Player.
It's not an Atomos or "how it was shot" issue. Otherwise I'd see the jittery-ness 100% of the time even in the original files. This is not a frame rate issue. This is a "playing the frames back in the wrong order" issue. And yes, I am aware of how 30fps clips in a 60fps timeline work. I'm dumb but I'm not THAT dumb.
I have this problem too. I'm using Mac Studio (M1 ULTRA). I got Jittery/Frame out of order (whatever it called) when "there is something to render in FCP". Like splitting video with 2 layer or even slow video with optical flow. It look normal when playback without render, but when I export it out, frame start to dance and swap like crazy.
I want to post some example, but I have to shoot some generic video to do it first.
I'm using all Prores 422 video to edit (shoot with RAW and convert to proxy in Davinci Resolve to Prores 422 for editing).
Can anyone know what happened? It's only happened on M1 ULTRA, it's okay on INTEL MAC.
I have to move back to INTEL everytime I have to export file.
...I'm using all Prores 422 video to edit (shoot with RAW and convert to proxy in Davinci Resolve to Prores 422 for editing)....
What is the original camera codec, the frame rate of the clip and the frame rate of the project in FCP?
If your original material was BRAW, on the Resolve media page, select the Metadata tab at upper right, then click the down arrow to the right of Media Pool and select "Camera" metadata. What was the clip fps at top of that pane, then below that what was the camera FPS and shutter speed? Blackmagic cameras have a separate sensor frame rate (aka "camera FPS") and project frame rate. It is very easy to put those cameras into HFR mode which can cause odd playback results.
In FCP using Inspector if you examine the converted ProRes 422 clip corresponding to the BRAW clip, what fps does FCP report?
On my M1 Ultra Mac Studio using FCP 10.6.3, I've edited and exported lots of UHD 4k/23.98 ProRes 422 material from an Atomos Shogun 7 captured from the Sony FX6 and so far have not seen this. I've also converted a lot of 6k 23.98 BRAW material from a BMPCC6K to ProRes 422 using EditReady and also haven't see this issue when editing on the M1 Ultra in FCP 10.6.3.