DNxHD encoding issues

[Vininski]

I hadn't noticed any issues in small scale tests I'd done with this software but now I am using it for a large project and I'm having a lot of issues with the DNxHD185X mediafiles it creates.

Out of around ~200 clips about ~5-6 of them will not show up in Avid when I drag the .mdb file into a bin. Additionally sometimes ~1-2 other clips will show in Avid and scrub through, but not playback in realtime instead Avid spits an error (regarding wrong number of frames, will add here when it comes up next). The weird part is, all of these clips work perfectly when run through the same workflow again. I noticed that with the clips that wouldn't playback, they have a different timecode when generated the second time.

I even ran a whole batch twice and noticed that different clips failed each time. All of these clips report success in the monitor. This makes it very hard to diagnose the issue.

Does anyone have any ideas?

[emcodem]

Hi Vin,
is it possible for you to turn on logging in the Workflow properties->special tab and upload one good and one bad log of the same file here?

cheers,
emcodem

[Vininski]

Thanks for looking into this emcodem.

I've just run another batch today and one of the clips (Clip0102) that transcoded even plays in Avid but the Tape and Start TC fields are blank. But lets look at the main issue I'm having first:

Clip0018 - Did not read in Avid on first transcode. Run process again and now it reads in Avid. Two logs attached.

Note: I have made minor changes to the workflow since my first post in an attempt to resolve the issue and also the Tape needed changing to avoid conflicts but the same issues are present. This is the first batch using it. I also had already tried running the workflow locally on one machine instead of the farm and it made no difference.

[Vininski]

Just found a clip with that frame issue, it popped up when I tried to read the audio timecode on all the clips in the bin. Here's the exact error message:

Error on clip "FRS3_SD05_A005_Clip0019":
Audio::PMM_INSUFFICIENT_MEDIA()
Player version of frame duration is longer than actual
[228] (228)

Error2.PNG (6.13 KiB) Viewed 10329 times

Clip0019 - Spits above error in Avid on first pass, It also occurs when trying to play the clip in the source monitor. Works fine on second pass. Timecode on both transcodes are different. First pass 10:30:24:09, second 10:30:33:12. Logs attached.

[admin]

Hi Vininski, thank you for using FFAStrans and welcome to the forum!

I'm having a hard time seeing what might cause issue in Avid. It might be the remuxing in bmxtranswrap that cause the problem. Unfortunately the current version of FFAStrans has limited logging so it's hard to see anything regaring this issue by just looking at the logs. So, the best thing would be to share the both the original and the faulty transcoded clips.

-steinar

[Vininski]

Thanks steinar, love the software BTW. It's exactly what I've been looking for.

I've already overwritten that file from earlier so here's another from the latest batch. Interestingly I've tried opening it in Resolve and it appears to be perfectly fine. Also I tried moving the mediafiles to another folder and pulling that mdb into a new bin and it appeared but with the wrong clipname and references different files from "FRS3_SD06_A007_Clip0002".

I've uploaded the first and second pass here:
https://drive.google.com/open?id=198O9_ ... Id0IhyC3VS

[emcodem]

Hey Vin,

thanks for the logs and the files, very helfpul!
I compared both of them and to be honest the files are so much the same that i don't think your "don't work" example shows any kind of defect.
To be honest the only real difference that i found was that one file was created on the machine "ASSIST-1" and the other on "ASSIST-2".

Tests done:
-) Compared essence (audio ch1 and video): md5 is the same (so the essence is really the same!)
-) Compared mxf header metadata dump: no difference besides the stuff that changes with every mxf: UIDs etc...

I guess the reason why the "don't work" example did not work is not related to the file structure.
Is it possible that the error is not related to the files that are produce by ffastrans but something like the avid import procedure or similar?

cheers!
emcodem

[Vininski]

Thanks again for looking into this. I'm puzzled as to why it's not working since I use the same workflow with Resolve to Avid all the time and haven't ever had this issue before which is why I had assumed it was a problem with ffastrans or specifically the workflow I created in it. Is there something unusual about how I'm importing these files?

1. Transcode files in ffastrans
2. Move folder of transcodes into Avid Mediafiles\MXF
3. Rename folder to "1" or similar
4. Open Avid window, let it scan and generate .mdb and .pmr files
5. Drag .mdb file into Avid bin, populates with clips
6. Rename folder back for neat organisation

I find it interesting that the md5 check you did was the same. I've done a "File Content" compare in FreeFileSync and it suggested the files were different, but I guess that could be the difference in the UIDs since AFAIK FFS does a bit by bit compare.

Just to add more info, I'm using:
Avid Media Composer 8.8.0
Windows 8.1 Pro

[emcodem]

Hm, maybe @momocampo can say more about the import procedure, i am not an Avid user/admin...

Can you tell if your "dont work" example does not work, even when you retry the import procedure multiple times?

When comparing the videos, you missed to extract the essence, mentioned in point 3:

What i do when comparing/analyzing mxf files:
1) manually investigate start and end in hex editor, check if files are complete
2) bmxtools: MXFDump.exe --validate-mxf inputfile.mxf
3) extract essence (bmxtools: mxf2raw -p "outputfile.raw" "inputfile.mxf")
4) investigate/compare essence using hex editor and byte -byte comparison/ md5
5) dump header metadata and indexes (bmxtools: mxfdump --show-dark --no-limit-entries --show-run-in --hexadecimal inputfile.mxf > inputfile_fulldump.txt)
6) compare the header metadata dump "manually" using diff function of some text editor (e.g. notepad ++)
7) manually check mediainfo advanced output of both files
8) check ffmpeg warnings/errors for "ffmpeg -i inputfile.mxf"
9) use IRT analyzer (only for OP1a)
10) user Baton QC (only for OP1a)

[momocampo]

Hello Vininski,
Well, I work with Avid too so I know well the different workflows.
I tried your first file (Clip1_dud) and it works well for me into avid (2018.2 in my case). This is DNxHD 185X.
Anyway, your workflow seems ok for me. I used each week a workflow to catch mxf file on ftp then convert it into atom before deliver into a Avid MediaFiles. It works perfectly and it is so useful !
Can you confirm me your "dud" file isn't ok for you?
Thanks

Benjamin

Edit : Well I tried the 2 files (dud and working) in Avid and both work well. Can you explain me what is your problem please? (sorry I read the previous posts and I don't exactly understand what is the problem (maybe also because English isn't my favourite language )