Panning/turning Mavic for slow smooth video capture - mid way stutter/pause
- Thread starter Grumpybeard
- Start date
-
- Tags
- mavic
Seeing that pan issue here, from my background in having done various realtime operating system app implementations
some stuff comes to mind. You have got some things going on in parallell;
- the camera panning in progress.
- the recording of video.
With little knowledge of the DJI design I assume that the camera module itself is a separate module with its own processor and operating system. As such it would be responsible for both (1) the motor drive that turns the camera, and (2) its video recording activity, storing recorded video on memory card.
What I assume is that after some seconds of video recording, with panning in progress, the camera reaches the end of a section of RAM memory buffer, and now does two things; (1) begins saving of that recorded portion to memory card, and (2) simultaneously continues the video recording in the presumable 2nd half of available RAM memory.
- What I assume happens is that the memory card buffering operation for some reason halts the camera pan motor operation while it saves to the card.
I would assume that the slow panning is an operating system task where camera internal software repeatedly sends stepper motor
commands to advance the panning. While that panning is ongoing a memory card write operation starts. My assumption is that the memory card write task has a higher operating system priority than the motor stepping task. This causes the motor panning to halt for the duration of the memory card buffer write job. When the buffering completes that process ends and is no longer having higher priority than the motor panning and its operation now continues.
This would be a software developer design error where the developer have not anticipated the camera buffering task, or more specifically this case of long term panning/recording might not have been a test case where the DJI camera supplier found a fault.
If I would guess again I would assume that the software that buffers video data to memory card is developed by a different developer than the one who developed the camera panning.. In the world view of the video buffering programmer nothing is more important than saving quickly to memory card ! So he grabs all priority he can get.
The developer of the camera panning stepper motor code is not aware of this. Both assume that their code is fine.
But my assumption is that the person who developed the camera buffering operation is inadvertently blocking out the camera panning task while he is saving data.
What you could do to check if my assumption is correct is to try using a different memory card with different speed characteristics.
If that card takes longer to buffer data you would see a longer camera pan pause. If it is a faster memory card the panning pause would be shorter (or just possible, not occurring). Worth a try. You would need to record for a long enough time while doing a slow pan just like you did in your example so that a camera buffer write would occur.
Again this is a suggestion of what might be happening.
I recently saw a similar bug in video cameras used on Tesla cars where the camera in that case loses some seconds of recording while buffering to card.
some stuff comes to mind. You have got some things going on in parallell;
- the camera panning in progress.
- the recording of video.
With little knowledge of the DJI design I assume that the camera module itself is a separate module with its own processor and operating system. As such it would be responsible for both (1) the motor drive that turns the camera, and (2) its video recording activity, storing recorded video on memory card.
What I assume is that after some seconds of video recording, with panning in progress, the camera reaches the end of a section of RAM memory buffer, and now does two things; (1) begins saving of that recorded portion to memory card, and (2) simultaneously continues the video recording in the presumable 2nd half of available RAM memory.
- What I assume happens is that the memory card buffering operation for some reason halts the camera pan motor operation while it saves to the card.
I would assume that the slow panning is an operating system task where camera internal software repeatedly sends stepper motor
commands to advance the panning. While that panning is ongoing a memory card write operation starts. My assumption is that the memory card write task has a higher operating system priority than the motor stepping task. This causes the motor panning to halt for the duration of the memory card buffer write job. When the buffering completes that process ends and is no longer having higher priority than the motor panning and its operation now continues.
This would be a software developer design error where the developer have not anticipated the camera buffering task, or more specifically this case of long term panning/recording might not have been a test case where the DJI camera supplier found a fault.
If I would guess again I would assume that the software that buffers video data to memory card is developed by a different developer than the one who developed the camera panning.. In the world view of the video buffering programmer nothing is more important than saving quickly to memory card ! So he grabs all priority he can get.
The developer of the camera panning stepper motor code is not aware of this. Both assume that their code is fine.
But my assumption is that the person who developed the camera buffering operation is inadvertently blocking out the camera panning task while he is saving data.
What you could do to check if my assumption is correct is to try using a different memory card with different speed characteristics.
If that card takes longer to buffer data you would see a longer camera pan pause. If it is a faster memory card the panning pause would be shorter (or just possible, not occurring). Worth a try. You would need to record for a long enough time while doing a slow pan just like you did in your example so that a camera buffer write would occur.
Again this is a suggestion of what might be happening.
I recently saw a similar bug in video cameras used on Tesla cars where the camera in that case loses some seconds of recording while buffering to card.
Last edited:
An interesting suggestion, but why did in not happen at all on the 3rd drone I was sent? Why does it happen at random intervals on a faiulty drone? It could do three short stutters in a row and then not stutter for a couple of minutes. If this was to do with card speed it would only vary if there were dramatic differences in the video content and therefore the compression levels possible.
Also. I'm pretty sure it happened when I wasn't recording, although in fairness I can't be sure of that. When I was running the tests I always ran the recording for evidence!
My current MP hasn't been out for many months (because of the virus and because I now have a Mini as well and it's easier to carry around.). It has shown the fault, but only on rare occasions. Next time I fly it, I will experiment. If the stutter only occurs while recording, then yes, this could be a coding rather than hardware issue.
Ok maybe I took imagination too far. I just read about it briefly and assumed it was occurring consistently. Good to hear they were able to send you a working one. That takes away worries for those thinking of getting one.
DJI Drone Deals
1. Mini 2
2. Mini 3 Pro
3. Mini 4 Pro
4. Air 2s
5. Air 3
6. Avata 2
7. Mavic 3 Pro
8. Mavic 3 Classic
2. Mini 3 Pro
3. Mini 4 Pro
4. Air 2s
5. Air 3
6. Avata 2
7. Mavic 3 Pro
8. Mavic 3 Classic
New Threads
-
-
What am I doing wrong? Unlock after LAANC approval.
- Started by twickers14
- Replies: 0
-
-
-
Forum statistics
