Welcome,
Guest
{JFBCLogin}
|
25 Jan 2021
New boarders will have their posts moderated - Don't worry if you cannot see your post immediately.
Read More...
Read More...
TOPIC:
library corruption 20 Feb 2023 04:58 #124276
I originally posted for help on the Apple Community, and one user offered to assist me in checking if there were issues with my library. However, there seems to be a security mechanism in place on the Apple Community that prevents me from successfully sharing the write-only file with her. This user suggested that I post on another forum to see if we can continue resolving my issue. So, this is a continuation of the discussion from the Apple Community.
* The original post on Apple Community: discussions.apple.com/thread/254642890?a...1192020#258690594022 |
|
Please Log in to join the conversation. |
library corruption 20 Feb 2023 15:09 #124285
You can upload your Library XML file or other diagnostic info to this secure write-only DropBox folder. I will examine it for any clues about the problem: www.dropbox.com/request/1EpOMd32gaTFl4ThhsZI |
|
Please Log in to join the conversation. |
library corruption 21 Feb 2023 02:04 #124303
OK, I have your fcpxml file and am examining it. Edit/Add: A quick evaluation shows no major problems. The library contains about 2,800 clips distributed across approximately 42 events, and about 84 hours of material. I don't have the media, just the XML. It seems to be mostly 1080p. However, on my M1 Ultra machine, it loads and runs OK, but (as expected) all clips are red and there is no media. Nonetheless that type of test can sometimes uncover problems. I don't see any obvious corruption in the project or library. There are several projects; the longest is about 2 hr. There are lots of rate-conformed clips that don't match the timeline frame rate. That can cause some computational overhead, but a higher-end machine should handle that. There are a total of about 3,202 keyframed audio adjustments across all clips. I don't have them broken down by project yet. There seem to be about 242 stabilized clips and 444 video effects, not all in one timeline; this is just a total from scanning the library XML. On a couple of the 2 hr timelines, those have 2453 IMG files as connected clips used as subtitles. If I'm reading it correctly, the XML shows there might be a huge number of markers, possibly over 7,000, across all clips and projects. We know having tons of markers will definitely slow down FCP. So my first suggestion is to reduce the markers (assuming that many actually exist). Another workaround might be to remove the main project to its own library. It can still connect to the same media files if using "import in place" media. |
|
Please Log in to join the conversation.
Last edit: by joema.
|
library corruption 21 Feb 2023 12:35 #124314
Thank you for your reply. I really appreciate your help these past few days.
I have now moved the main editing project to a new library for editing, and also deleted unnecessary markers, hoping to see some improvement. Regarding the frame rate issue, I'm not sure if the frame rate of the earlier footage is different from the later footage, so I'm unable to change the timeline frame rate. Once again, thank you for your assistance these past few days. |
|
Please Log in to join the conversation. |
library corruption 21 Feb 2023 19:47 #124321
If the clip frame rate differs from the timeline frame rate, FCP will automatically conform that to the timeline. It's not a big compute burden unless you use optical flow rate conforming (which I don't think you did). Further study of your library and project XMLs shows a huge number of markers which for some reason do not appear on the timeline or in the timeline index. E.g, just in the project "rough cut v2 (Snapshot of v1)", the project XML shows 1,229 markers. Yet within FCP the timeline index of that same project shows only 54 markers. Maybe they are within compound clips and show in the XML but not in the timeline unless you open each compound clip. Many of your markers have traditional Chinese names, so maybe it's some kind of bug related to fonts or localization. As a test I recommend you select a project which has consistent problematic behavior, duplicate that project, open it then delete all markers. If there are lots of markers in compound clips you might need to delete those as part of the test. Restart FCP then open the project and see if that works any smoother. We know from experience that FCP may exhibit performance problems if many thousands of markers are used. I don't understand the disparity between the number of markers shown within FCP vs the number listed in the XML. Maybe the markers are leftover from some FCP crash on an earlier version of the library. To investigate this you can export a project XML, in Finder open the package, the rename the info.fcpxml file to be YourFileName.txt, then examine the file in TextEdit or other similar text editor such as BBEdit. If you search on the phrase "marker start" you will see each marker. They are denoted this way: <marker start="8649641/60000s" duration="100/3000s" value="Marker 4"/> The marker start item is the clip timecode of that marker start location where the quotient is expressed as fractional seconds. You can see the corresponding clip timecode in FCP by clip skimming over that clip with the timecode readout set to seconds (Final Cut Pro>Settings>General>Time Display>Seconds). In the above case the marker start clip timecode is 8649641/60000 = 144.1606 seconds. Note this is clip timecode, not sequence or timeline timecode. The duration is a quotient indicating typically one frame (for video) or one sample (for audio). The marker name is listed after that, in this case it's called "Marker 4". The difficult part is finding the clip to skim to look for that marker by timecode. Normally in the XML file the clip name is stated a few lines above that but for compound clips, I don't know how that's handled. You might find it easier or useful to use a tool like grep on your project XML file and extract the lines containing "marker start". That will include a line for each marker. You might be able to deduce something from reviewing the marker names. You might conclude "I wondered what happened to those, they've been missing". Or maybe you will conclude "Yes I knew about those, they should all be there". To use the command-line version of grep (which is included in the MacOS terminal), it would be something like this: grep "marker start" RenamedFilenameOfXML.txt > OutputFileToContainedExtractedLines.txt Or you could use a GUI version of grep such as Visual Grep: apps.apple.com/us/app/visualgrep/id597887722?mt=12 I currently think the main problem might be the tremendous number of markers (assuming the XML is correct). It shows 7,625 markers in the library. In the project Rough Cut V1 are 1,100 markers, in Rough Cut V3 there are 1,207 markers, likewise in Rough Cut V4. There could be other issues but we know very large numbers of markers will cause problems, so I'd suggest focusing on that as a troubleshooting step, then evaluate the performance and stability. |
|
Please Log in to join the conversation.
Last edit: by joema.
|
library corruption 23 Feb 2023 17:44 #124349
Hi, I have tried for a long time and finally completed the grep operation you mentioned.
I tried to delete the markers of Rough Cut 4 in the new library, but I only deleted the visible 117, and the remaining 1089 were hidden and could not be found in the new library. These hidden markers are in the old library. Would you recommend how I can delete these large numbers of markers next? Should I import the original library XML, delete all markers, and then export a new XML? |
|
Please Log in to join the conversation. |
library corruption 24 Feb 2023 08:50 #124357
Update: I tried importing the original library XML and attempted to delete all markers. However, FCP runs very slowly and sometimes takes over 30 minutes to open one of events with footage. Is there a way to directly delete the remaining 1089 hidden markers from Rough Cut V4 in the new library that only contain the project 'Rough cut v4'?
|
|
Please Log in to join the conversation. |
library corruption 10 Mar 2023 23:08 #124551
I'm sorry about my delayed response, I was on a field assignment and am getting caught up. It is theoretically possible to delete all markers from a project XML file, then delete and reload the project from the edited XML. This is an experimental procedure so make sure you have an FCP duplicate of that project, and everything backed up, especially the library. Have a file-level backup by duplicating the library bundle and a library XML exported. Also export a copy of the project XML you will experiment on. Also save all copies of the FCP auto backups of that library which are by default in /Movies/Final Cut Backups. I tested this on a copy of your library and it seemed to work but there are no guarantees. Procedure: 1. Export a project XML of the project you want to delete markers from (in addition to a project XML for backup which you do not touch). 2. In Finder do right-click>Show Package Contents on the .fcpxmld package (the project XML you exported). 3. In terminal use grep or Visual Grep or BBEdit if you have that. Below is the syntax for grep in terminal: grep -v "marker start" input_file.fcpxml > output_file.fcpxml Where: input_file is the existing Info.fcpxml file and out_file.fcpxml is a new filename you provide. 4. After that is done you can delete the Info.fcpxml file and rename your new edited file to Info.fcpxml. 5. In FCP delete the project and try to reload it from the edited XML. In my test this worked, deleted all hidden and non-hidden markers and had no apparent side effects. However I don't have your actual data, just the library and XML files. You should be prepared to recover from backups. I only cautiously suggest these steps because of your difficult situation. If you have the 3rd-party utility Beyond Compare, it can compare two different versions of XML files and show you how they differ. E.g, after using grep to delete all lines in the XML which contain "marker start", you can use Beyond Compare to verify those were the only changes made vs the backup XML copy. Beyond Compare: https://www.scootersoftware.com Besides the many thousands of markers, there is another issue which might be impacting your performance. That is having very many compound clips, especially compound clips which contain audio in A/V files (vs a separate audio file). I am investigating that problem. Therefore another theoretical workaround might be to detach the audio from your A/V clips. I don't know if that is possible for you. You might duplicate a project and test it on a small scale first. An easy and safe step that might help some is selecting the "labels only" timeline view. It takes less UI overhead to maintain that. That is CTRL+OPT+6, or you can use the filmstrip button at bottom right. |
|
Please Log in to join the conversation.
Last edit: by joema.
|