Video glitch/stutter #1855
Replies: 1 comment 2 replies
-
|
Does it occur when you make movies with the start/stop event? |
Beta Was this translation helpful? Give feedback.
-
|
You mean manually via the web control? |
Beta Was this translation helpful? Give feedback.
-
|
I recorded several short (~10 sec) videos using the web interface (Start Event/End Event). These seem OK. |
Beta Was this translation helpful? Give feedback.
-
I have the occasional video exhibit this issue. A moving object will just enter the side of the field of view (either R to L or L to R) and the video will freeze for several seconds. Then it will move normally for a few frames, only to have the object disappear somewhere in the middle of the field.
In researching, I found this issue and saw the comment about moving the output folder to /tmp to avoid slow I/O to the μSD card. That made good sense; however, I do not think /tmp is kept in RAM. So I created a folder on tmpfs. Testing it with dd, writing a large file is over 20x faster than to a regular folder on the card. I allocated 50MB, which is twice the largest video I have seen.
I also modified my on_movie_end script to move the video from the tmpfs directory to a regular directory on the μSD card.
As luck would have it, after making these changes the second video recorded exhibited this issue (1-20250303-080955-001.mkv in the log, and also 1-20250303-082320-005.mkv) so the changes do not seem to have had an effect. I hope someone can offer another idea.
The machine does not seem to be running hard. I occasionally will see a one-minute load average slightly over 1.00, which seems fine for a quad-core CPU.
Environment:
Motion 5.0.0-git-20250224-6e685d8 (built from source)
Pi camera v2
Debian GNU/Linux 12 (bookworm) aarch64 (Raspberry Pi OS Lite 2024-11-19)
Raspberry Pi Zero 2 W Rev 1.0
Linux cam1 6.6.74+rpt-rpi-v8 #1 SMP PREEMPT Debian 1:6.6.74-1+rpt1 (2025-01-27) aarch64 GNU/Linux
Beta Was this translation helpful? Give feedback.
All reactions