After adding a source video (around 11 minutes) which is 3600x576, the GPU usage is around 25% and the CPU efficiency cores are at about 17%. This remains the same even though the TVAI UI is “idle”. If I minimise the app, the GPU drops to the usual 1% or so but the CPU usage remains the same. If the only input source is a more usual aspect ratio video, the CPU / GPU usage is the expected minimum.
This issue of high CPU/GPU usage when idle doesn’t occur with v5.5.
(The video is a horizontal stack of 5 x 720x576 segments which is part of my work-around for slow SD performance on Apple Silicon, in case anyone is wondering)
Also, I’ve just checked, and this also happens in v 6.03 if the video is a “standard” 1920x800, but not with 1920x1080. It appears to be a problem with generating the thumbnails as these aren’t appearing in the case of the wider aspect ratios.
Since I didn’t have such a widescreen vid, I used Handbrake to stretch one to a size of 1436x280 (anamorphic), loading this one into TVAI I suddenly got this:
And that’s just loading the video file into TVAI without any processing.
So, yes there’s sudden high GPU usage when there shouldn’t and failure of the thumbnails creation as a likely reason (stuck process?)
Just for completeness, I can confirm this bug is also in the Windows version too, for wider aspect ratios, and hence also fails to show the thumbnails.
It’s a very basic Windows 10 machine. I don’t usually use Windows for TVAI but it was just to verify if this bug was also in the Windows version, as I thought that would be helpful.
Given this is so easy to replicate, and with additional input from user jo.vo, I trust you now have everything you need.
[Parsed_crop_1 @ 0x600000155d90] Invalid too big or non positive size for width ‘100’ or height ‘50’
[Parsed_crop_1 @ 0x600000155d90] Failed to configure input pad on Parsed_crop_1
[vf#0:0 @ 0x600000754dc0] Error reinitializing filters!
[vf#0:0 @ 0x600000754dc0] Task finished with error code: -22 (Invalid argument)
[vf#0:0 @ 0x600000754dc0] Terminating thread with return code -22 (Invalid argument)
[vost#0:0/png @ 0x13a607fc0] Could not open encoder before EOF
[vost#0:0/png @ 0x13a607fc0] Task finished with error code: -22 (Invalid argument)
[vost#0:0/png @ 0x13a607fc0] Terminating thread with return code -22 (Invalid argument)
[out#0/image2 @ 0x600000354540] Nothing was written into output file, because at least one of its streams received no packets.
frame= 0 fps=0.0 q=0.0 Lsize= 0KiB time=N/A bitrate=N/A speed=N/A
Conversion failed!
Word from the team is this is the preview thumbnails being generated for the section of timeline being displayed, once those are handled the usage will throttle back off.
Just that as @Chamberhouse reported those (thumbnails) are never handled / finished with this type of video.
So, again:
With such extreme widescreen aspect videos the thumbnail creation fails, likely some process gets stuck which is then leading to that permanent GPU and CPU usage.
Andy can you share the input file you are using for the team to investigate further? I sent the snippet of your logs over to them.
You can securely submit your file(s) to our Dropbox using the link below. Please be sure to send me a note to let me know you sent something and include the file name.
As you can see there are no thumbnails with this extreme wide format video (see the empty timeline) and high GPU/CPU usage - and this stays like that forever.
With a “normal” aspect video TVAI behaves as it should:
thumbnails are there and CPU/GPU is mostly idling.
And this issue (while only occurring in such a quite seldomly used setting) still really is extremely easy to reproduce as it persists over at least the last 6.x versions of TVAI and occurs independent of the system / on both, Mac and Windows.
But if you want, I can provide you with the vid I tested with.
Hi Kyle, I uploaded a very small sample file “Mini-Stack.mp4” which is sufficient to replicate the issue. Any video with aspect wider than 2:1 would also do.