Left Repeater Video Corruption After 2019.32.2.2

[DCEV]

Has anyone seen issues with video corruption of one of the repeaters after 2019.32.2.2?

I pulled up both Sentry and TeslaCam footage and unfortunately I'm seeing consistently sporadic video corruption on one of the repeaters after 2019.32.2.2.

I have already done the following to try and fix the issue to no avail:

• Rebooted car.
• Delete all files in the TeslaCam folder.
• A brand new high speed USB drive FAT32 formatted with the TeslaCam folder.

None of these have solved the data corruption issue with one of the repeater cameras. Anyone have this issue or have had this issue in the past? What solutions can I try? Up until now, TeslaCam and Sentry videos have been reliably recording left, right, and center cameras.

I really need Sentry and TeslaCam to reliably record videos.

What really sucks about this situation is that if there are recording errors or a camera is not recording or recording with corrupted data, you'd have no idea, until you actually need the footage. Wish as the video files are being written, if there is an issue, the DashCam icon would include an exclamation mark so you know something needs to be fixed.

Here you can see how the left repeater camera has odd file sizes with corrupted video:

And this is what one of the left repeater corrupted videos look like:

 

[cRiordon]

Our Model 3SR+ built in 03/19 has had left repeater corruption issues since the inception of Sentry Mode. Rare or non-existent when the car is in motion, but very frequent when the car is parked. Pretty annoying, especially if there is a Sentry event. It very rarely has happened with the right repeater, and never with the front camera. I thought it might be a hardware or connection issue, but an analyst tech who is working with me on a couple of other issues which the car came with said it is a known issue and they are working on an updates to resolve it.

Still running on 2019.28.* in part because after nearly 6 months of ownership, the car still does not have any wifi capability. Mobile Service replaced the antenna a couple of months ago, and are now apparently waiting for a couple of cables and a computer to definitely resolve the issue.

 

[DCEV]

Could you all check your Sentry/TeslaCam videos to see if everything looks good? I would never have known about the corrupted video from one of the cameras unless I checked.

Anyone else have issues with video corruption for one of the repeater cameras?

 

[DCEV]

No one else is having issues with corrupt video recordings on certain channels with TeslaCam and Sentry?

 

[MNoICE618]

Yes I am getting this issue too since sentry started working in Australia.

I wonder if this corruption causes the “dancing cars” on the dash surrounding vehicles image.

 

[DCEV]

Yes I am getting this issue too since sentry started working in Australia.

I wonder if this corruption causes the “dancing cars” on the dash surrounding vehicles image.

Dancing cars just has to do with how the camera data is currently processed by the NN and is unrelated. The dancing will eventually subside and stop as the NN is perfected.

This issue just has to do with Sentry and Dashcam footage of one of the repeaters being consistently corrupted.

 

[MNoICE618]

Dancing cars just has to do with how the camera data is currently processed by the NN and is unrelated. The dancing will eventually subside and stop as the NN is perfected.

This issue just has to do with Sentry and Dashcam footage of one of the repeaters being consistently corrupted.

DCEV, that is an interesting way to shut down a conversation; please provide your source of information and background; an experienced automotive engineer apparently, so I'd like to know more.

I work with embedded processor signal processing for sensor systems which feed the ADAS for several Tier-1 automotive suppliers, so I have considerable experience with this type of system; and to me, that corrupted video sensor input is not a friend of object recognition.

If one reviews the car's saved driving video, it is evident the corruption is not solely a Sentry issue.

Just saying.

 

[DCEV]

DCEV, that is an interesting way to shut down a conversation; please provide your source of information and background; an experienced automotive engineer apparently, so I'd like to know more.

I work with embedded processor signal processing for sensor systems which feed the ADAS for several Tier-1 automotive suppliers, so I have considerable experience with this type of system; and to me, that corrupted video sensor input is not a friend of object recognition.

If one reviews the car's saved driving video, it is evident the corruption is not solely a Sentry issue.

Just saying.

Didn't mean to be disrespectful or shutdown the conversation but from the dancing cars it looks like the NN is currently processing individual stills without temporal smoothing and this is why the position of vehicles in close proximity change so widely. I feel especially with HW 3.0, with better software and processing capabilities, there will be more temporal context and continuity in the rendition of the synthetic model of what surrounds the car.

Also I feel what causes video corruption on one of the repeaters is limited to what is saved by TeslaCam or Sentry as otherwise I am sure a non-functioning camera would cause what an Autopilot self-test routine to fail before allowing Autopilot to be engaged.

While video from a repeater is corrupted, Autopilot works as usual so this is my rationale for thinking this is just something to do with a glitch with processing video from one of the repeaters.

So now one else is having repeater video corruption issues? Have you all looked at your Sentry and TeslaCam footage recently?