shift video to audio synch with another source

[poisondeathray]

ffmpeg seems ok too and doesn't crash

ffmpeg -i "test_zero_frames.avs" -c:v rawvideo -an -f null NUL

 [out#0/null @ 0000004d0260cbc0] Output file is empty, nothing was encoded(check-ss / -t / -frames parameters if used)
frame=    0 fps=0.0 q=0.0 Lsize=N/A time=N/A bitrate=N/A speed=N/A

Passing zero length video and audio (no -an) is ok too

[gispos]

I don't know if it's just about looking? If so, that's what I do.

Split_View_Offset

[buddha9]

Originally Posted by gispos

I don't know if it's just about looking? If so, that's what I do.

Split_View_Offset

Inmpressive!
Thanks so much gispos!
Can you give me this script please?

[gispos]

Originally Posted by buddha9

Inmpressive!
Thanks so much gispos!
Can you give me this script please?

This is not a script, it works in AvsPmod with SplitView and Video > Tools > 'Locate frame'.
A group must be set for the current tab and the tab to the right of it. Then start SplitView in the left script and start Video > Tools > 'Locate frame'. The non selected script is frozen with 'Freeze frame'.

To find the offset automatically with 'Locate frame', the wonderful 'LocateFrames' function from Doom9 member StainlessS is needed.
https://www.mediafire.com/folder/hb2...7z6/StainlessS or http://www.sendspace.com/folder/2mwrco and you can also read Help > 'Locate frame readme'

However, you have to limit the search radius with start and stop, otherwise the search will take too long. So manually scroll in the right script to close to the same position as the left script and then click on 'Start'.
The threshold may need to be set high (50) if the two videos are visually very different.

[FranceBB]

Originally Posted by poisondeathray

ffmpeg seems ok too and doesn't crash

Not any longer, but it used to. I remember the conversation StainlessS pointed at very well: Link
It was April 2nd 2023 and during the night someone started uploading several images via Aspera as they were recording a sailing competition in New Zealand and they also shot some pictures. Those got downloaded automatically, then they were indexed, converted with all the various adjustments, a mute audio track was inserted and they were eventually supposed to be re-encoded to XDCAM-50 (i.e MPEG-2 50 Mbit/s + PCM 24bit 48000Hz audio muxed in mxf) and checked in into Interplay automatically. Unfortunately, with images being "a picture" they resulted in one single frame called frame 0, so a BlankClip(length=0) was added to create the mute audio track but that turned out to be a very bad choice 'cause what FFMpeg did was reading the script and it just kept going... and going... and going... and going... forever. Quickly enough all the nodes in the farm ended up being saturated and overwhelmed and the whole thing crashed. Restarting didn't help as it started processing the next batch of images until it crashed again. I even saw one server with Windows Server 2019 Standard x64 and 128 GB of RAM reaching 1 TB of paging file before giving up and crashing.

The short term solution was forcing BlankClip() to use a length that started from at least 1, but it took me a bit to figure out the issue, change the code and re-compile.
BlankClip was using the $i_src_vid_frames variable and sure enough I added

If $i_src_vid_frames = 0 Then $i_src_vid_frames = 1

and it started working again.
Anyway, I'm glad to see that it has been fixed in FFMpeg as well now.