Upcoming integrated dashcam feature... How?

[insaneoctane]

Here's the thing, my Model 3 has narcolepsy. The thing is literally ALWAYS sleeping. I know because I do a lot of API monitoring when it's awake, which is like only when I drive it and a few minutes afterwards... Don't get me wrong, I think it is a good thing, because it minimizes vampire loses. I don't know how Tesla will offer the dashcam feature without giving back many of the recent vampire loss squashing gains...

 

[tracksyde]

I imagine if Tesla were to implement a dashcam feature, its going to be relatively featureless.. like no parking mode, not too much recording time (maybe a few seconds before a collision), not 1080P, etc.

 

[davedavedave]

Hmm... there’s dash cam in motion (I.e., recording in case of accidents) and there’s stationary dash-cam (I.e., in the case of vandalism or breakins). Perhaps we’ll only get in-motion dash-cam? Or perhaps the ultrasonic sensors are on a separate, lower drain power circuit and activate the recording for parked dash-cam capability?

Shrug - we will see....

 

[pavankp]

Hmm... there’s dash cam in motion (I.e., recording in case of accidents) and there’s stationary dash-cam (I.e., in the case of vandalism or breakins). Perhaps we’ll only get in-motion dash-cam? Or perhaps the ultrasonic sensors are on a separate, lower drain power circuit and activate the recording for parked dash-cam capability?

Shrug - we will see....

For parked dashcam to be useful, not only the ultrasonic sensors but also the cameras need to be always on and recording. Otherwise you will only get video of stuff after a sound has been detected. You'd want video of a few seconds before the sound as well.

 

[HankLloydRight]

Nothing Tesla does in this area will ever come close to buying and using a real dashcam. They shouldn't even waste their time on it, other than maybe adding a few +12v accessory wiring harnesses in convenient places for us to use. Two up near the rear view mirror, a few more in the back, one or two in the frunk area, etc. Then just let us buy the best dedicated "uni-tasker" device out there for our needs. If Tesla still can't fix the media player up to 2001 standards, I'd hate to see what their dashcam software would look like.

 

[Uncle Paul]

They could simply allow us to hook into a useful camera and use our own memory sticks to do the recording on a loop.

 

[SDKoala]

Nothing Tesla does in this area will ever come close to buying and using a real dashcam. They shouldn't even waste their time on it, other than maybe adding a few +12v accessory wiring harnesses in convenient places for us to use. Two up near the rear view mirror, a few more in the back, one or two in the frunk area, etc. Then just let us buy the best dedicated "uni-tasker" device out there for our needs. If Tesla still can't fix the media player up to 2001 standards, I'd hate to see what their dashcam software would look like.

I agree. All I really want is a constant source of 12V power that stays on when the car is unoccupied that's not going to break something potentially. The power requirement of a dash cam is so insignificant compared to the energy it takes to make the car run that it should be really easy to engineer.

There's a thread somewhere that shows images from the built-in cams on a Model S and they don't look very good. Dash cams record in 4K now. I read that the built-in cams are 720p. Good enough for EAP apparently, but the level of detail that you may want in a collision or act of vandalism.

 

[HankLloydRight]

Yup. And these days, the standard minimum feature set for dashcam software is very high, with motion sensors, parking modes, event recording, toggleable sound recording (for legal reasons), cloud uploading, built-in wifi access to the clips, and all kinds of settings for each of those features... just look at the mobile apps provided with the Blackview or Blacksys cameras... If/when Tesla releases a 'dashcam' feature, it will be better then not having any dashcam in the event of a significant accident, but it's going to be really bare-bones, and most everyone will still want to install a real dashcam anyway.

 

[zosoisnotaword]

Otherwise you will only get video of stuff after a sound has been detected.

I can't tell if you're serious or joking. Please advise.

 

[Knightshade]

I guess I don't understand how it's impossible for Teslas feature to be as good as a 3rd party dash cam.

The car has a ton of cameras and sensors.

More than a single dash cam would have.

The idea it's asleep most of the time to avoid drain is a non-issue. A mean- your third party camera ALSO drains power, doesn't it? So the tesla put into "dashcam on" mode would too, and just like the drain from the 3rd party one, it means you'd lose some range to power it.

What magic parts do third party cams have that the Tesla built-in stuff wouldn't?

Storage limits are the only thing I could think of- (maybe not even then since we have no idea how much storage the built in computer has, do we? But even that'd be easy to fix by enabling USB storage.

 

[HankLloydRight]

What magic parts do third party cams have that the Tesla built-in stuff wouldn't?

As already mentioned, the AP cameras have lower resolution and different color filters and frame rates to optimize machine vision processing and autopilot functionality. Exactly NOT what you need or want in a dashcam.

I guess I don't understand how it's impossible for Teslas feature to be as good as a 3rd party dash cam

It's not about the "parts".. it's about the software. Since Tesla has to start from absolute scratch to create "dashcam" software, it will never be close to what current dashcam makers currently provide, and Tesla will always be way behind as real dashcam makers continue to innovate.

It's like you're saying "I guess I don't understand how it's impossible for a Ford EV to be as good as a Tesla EV."

Tesla is the Ford of dashcams.

(and media players, but that's off-topic).

 

[Knightshade]

As already mentioned, the AP cameras have lower resolution and different color filters and frame rates to optimize machine vision processing and autopilot functionality. Exactly NOT what you need or want in a dashcam.

I guess that depends why you want or need the dash cam.

Though if it can use ALL the cameras that'd give it some angle advantages over any 3rd party single camera setup

It's not about the "parts".. it's about the software. Since Tesla has to start from absolute scratch to create "dashcam" software, it will never be close to what current dashcam makers currently provide, and Tesla will always be way behind as real dashcam makers continue to innovate.

Why?

dashcam software isn't especially complex.

It's a LOT less complex than, say, "making a car that can drive itself"

So the idea they can NEVER catch up on... dashcam software...is... hilarious.


[QUOTE="HankLloydRight, post: 2792140, member: 20679"
It's like you're saying "I guess I don't understand how it's impossible for a Ford EV to be as good as a Tesla EV."

Tesla is the Ford of dashcams.

(and media players, but that's off-topic).[/QUOTE]


... but that doesn't make sense either.

It's NOT impossible for Ford to make an EV as good as a Tesla.

It might be impossible for them to do it tomorrow. But if they wanted one in 5 years that'd be entirely possible.

They don't seem especially interested in doing so- but there's no reason at all that it'd be impossible

Tesla cars aren't magic. Neither are dash cams (and dash cams are WAY less complex than cars)

They are engineering and technology.

Which are both solvable problems with enough money and talent.

(and not likely all that much of either for the dashcam issue- especially when the car ALREADY has cameras built in)

 

[drawfour]

As already mentioned, the AP cameras have lower resolution and different color filters and frame rates to optimize machine vision processing and autopilot functionality. Exactly NOT what you need or want in a dashcam.

But are those characteristics of the hardware of the camera itself, or of the software processing? If the camera is capable of getting 30fps, but the vision software only wants 15fps, there's no reason they can't skip every other frame after a splitter. If they want 20fps though, then that's harder because 30fps won't easily split into 20fps and there would not be a constant flow of frames the right distance apart. Color filters could also be done in software after capture. Resolution though, you're not going to be able to upscale.

 

[HankLloydRight]

It might be impossible for them to do it tomorrow. But if they wanted one in 5 years that'd be entirely possible.

Yes, and Tesla will still be five years ahead of Ford.

And while it's not 'self driving' dashcam software is very complex.

There's no need to argue the point... the future developments will decide the path and we'll see who is more accurate at predicting it.

 

[Knightshade]

Yes, and Tesla will still be five years ahead of Ford.

If you assume every company makes exactly the same amount of progress at exactly the same rate over exactly the same time.

Which is completely untrue.

Otherwise every company first to market would remain the leader of that market, always. Which simply isn't how the world actually works.

See, for example-

IBMs spot in the PC market
3dfx spot in the 3D graphics card/gaming market
Ataris position in the home video game console market
Seahan/Nomad in the portable music player market
Motorola in the cell phone market (at least this one they're still IN the market- though far from the leader)

and on and on thought out the entire history of technology.

And while it's not 'self driving' dashcam software is very complex.

Not really, no. Especially compared to self driving.

There's no need to argue the point... the future developments will decide the path and we'll see who is more accurate at predicting it.

that's true- which is why throwing around absolutes and impossibles seems... ill advised.

 

[HankLloydRight]

Not really, no. .

Yes, it is. Just because it's not as complex as AP, doesn't mean it's not complex. It's not as simple as connecting a camera to a memory card. Have you actually used a current, advanced dashcam?

Tesla has a horrendous track-record of writing user-driven software like this (i.e. browser, media player, etc). They might excel at battery management, autonomous driving, and vision processing, but those are completely different skill sets.

If/when Tesla releases their 'dashcam' software, you'll see just how bare-bones and hard to use it will be compared to current, advanced dashcams on the market. The Tesla dashcam will be nothing more than a side-project distraction for a few entry-level developers just so Elon can say he did it. Tesla has real problems to solve and should just let the dashcam makers sell dashcams.

 

[Glamisduner]

Dash cams are not allowed at my work. Who makes a good one that can be removed and stashed in a couple seconds? Simply unplugging it is not good enough. My old one was pretty easy, but quality isn't the best compared to more modern cameras. I used it with a rear view mirror mount that is sorta locked into.

 

[jkennebeck]

I'd be ok if they implemented it with power to whatever systems needed to detect vandals AND crashes. They can put a disclaimer that says when main battery gets under x% these features will turn off. I don't think I've ever gone under 8% battery and I don't think the camera resolution would need to be 4K to see what happened. Maybe catch a plate etc.

 

[Knightshade]

Yes, it is. Just because it's not as complex as AP, doesn't mean it's not complex. It's not as simple as connecting a camera to a memory card. Have you actually used a current, advanced dashcam?

Yes, and it's not that hard.

Or "advanced"

Disclaimer- I'm a software engineer- so my perspective might be different than, say, carpenter, regarding how advanced any given type of software is.

I'm not really sure what features you specifically are thinking of that are "advanced" or at all difficult to program though- can you be more specific?

Tesla has a horrendous track-record of writing user-driven software like this (i.e. browser, media player, etc). They might excel at battery management, autonomous driving, and vision processing, but those are completely different skill sets.

You're talking about UI, not actual functionality of the dashcam.

Those are different things.

(often handled by different teams too)

Quite obviously they already HAVE software to use the cameras, and record footage from them, and to detect a ton of data via sensors. They'd just need to tweak that to handle recording/saving data a bit differently, and how they want to handle power/recording when the car is "off"

So when I tell you the back-end part is not "advanced" compared to what they can already do- it's pretty clear that's accurate.

The UI part- that indeed probably won't be super fancy out of the gate. But that doesn't impact its ability to do the fundamental job of a dashcam- record pictures and video of events for which you want a video or picture record.

Again I'm wondering what features, specifically, you think are so vital and advanced that it would lack?

 

[HankLloydRight]

I'd be ok if they implemented it with power to whatever systems needed to detect vandals AND crashes. They can put a disclaimer that says when main battery gets under x% these features will turn off. I don't think I've ever gone under 8% battery and I don't think the camera resolution would need to be 4K to see what happened. Maybe catch a plate etc.

Power consumption is really a non-issue. These things draw so little power it's pretty much negligible. I have a two-channel dashcam, a front camera switch (see: Front/Rear Camera Touchscreen Switch Kit - SOLD), and my pedestrian EV-horn all running 24x7 and have never seen any measurable decrease in range. My EV-Horn only draws 77ma on average, and that includes periodically recharging the battery in the USB speaker, which equates to less than one mile of power consumption per month.

and I don't think the camera resolution would need to be 4K to see what happened. Maybe catch a plate etc.

Yes, you definitely don't need 4k to catch an accident or vandal, but you need at least 1080 to catch a plate.