Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

FSD Beta Videos (and questions for FSD Beta drivers)

This site may earn commission on affiliate links.
Just stealing a clip with no credit or link to the original author is not fair use. It's like a lot of clippers that earn revenue just by stealing content by clipping other videos, but usually even they link back to the original author or they do something to make it "transformative" by editing in some way (or maybe adding translations, but even translations don't give you a free pass), not just post a single clip outright.
I would characterize the use of the clip as criticism and commentary.
 
I would characterize the use of the clip as criticism and commentary.
Criticism/commentary doesn't absolve you of at minimum giving credit and also making it "transformative" (your commentary being the substantial part of the given piece, not the clip itself). This was not the case in the original twitter post that exploded, although subsequent reposts may be closer to that standard.
 
Criticism/commentary doesn't absolve you of at minimum giving credit and also making it "transformative" (your commentary being the substantial part of the given piece, not the clip itself). This was not the case in the original twitter post that exploded, although subsequent reposts may be closer to that standard.
No one else seemed confused that the intent of the tweet was reporting on and criticizing Tesla's FSD testing program.
Where is there a requirement to give credit?
 
Though it does look like he came close to the girl, it’s somewhat surprising that no one pointed how irresponsible she was by opening her door into the path of an oncoming car. Ultimately he is responsible as driver in control. But what a reckless action from her. I would have stopped and given her a piece of my mind.
 
I think you have that backwards.
Wide angle cameras make things look further away than they are. Therefore the lady is closer than she appears to be, and the gap is narrower than it appears to be.

View attachment 711500
From Canon’s website:

Fine, you're correct in pointing out that wide angle distorts distances and "amplifies distance between objects", ie: objects appear further away than they are. Which makes distance calculations difficult in a distorted image.

Here is an overhead shot with minimal distortion from Google Earth. Approx: 522 N. Ada, Chicago. Example cars placed as best as I can estimate from the video (colours of car irrelevant, they are just cars I grabbed from the satellite image).. Tesla closer to curb car than double-parked car. Google Earth ruler for street width=36'.

Distance between Tesla and Lady's car ~3.6'. Distance between Tesla and open door ~1.3'.

Door2.png

YMMV
 
Last edited:
How do you know all the metrics they programmed in to look for and data mined to come up with this? Could you provide us with a document or link of what Tesla, or any insurance company for that matter is doing, to determine their rates and what constitutes safe driving?

Various hackers have published info from either the car computer code, or the app code, that's how we know.


The plan is for the app to eventually show you a safety score based on your behavior, and code is already there- but obviously hasn't gone live yet.


Already posted a link (I think in the post you are replying to) discussing some of what other insurance companies are doing on driver monitoring, though I'm unaware of anyone having hacked into their devices for deeper dives, not that I've had any reason to really look)
 
  • Like
Reactions: Nakk
He was in mid-sentence saying "Tesla lets us share everything on social" then he got cut off as the car was cutting into a lane that was not actually a lane (I guess this is why these shouldn't be streams, but commentary should be narrated separately), then he says "Tesla doesn't want us sharing all of the clips of the videos, um just like when it looks good because they know people take it out of context". So it's a bit self contradictory. And we have seen all sorts of videos where the system does badly, so not really sure what he really means by that. Maybe Tesla doesn't want individual clips, but is ok if you share the whole trip.
Its not contradictory, it makes 100% sense when you are trying to control and dictate the narrative. An average FSD beta video is what? 15-25 mins or more?
No one has that time and you lose the gravity of certain failures in the noise of a 30 mins video.

But no its not contradictory. sharing everything is talking about things. This is why he said "social"
For example if someone got into a serious accident with FSD beta today, saying "I got into a accident with FSD beta today" is different than posting the clip of the accident. That's what that means.

There's a reason why Tesla doesn't allow live streaming and its obviously not because of just attention. Tesla already knows its customers do live commentary, they talk to the camera, turn around, explain things, etc.

Its because if there's an accident they can detect it and reach out to the driver immediately and ask them not to release the video because "ppl will use it to hurt Tesla, blah blah blah". And tesla fans being blind loyalists will gladly follow along as proven by Gali.

That would be impossible to stop if things were live-streamed because the internet never forgets. It would be disseminated in seconds... Tesla knows that words have very little weight and then you have a picture that is worth a thousand words, but a video clip will get millions of views. As proven by the DMCA clip getting over 5 million views in about a day. Being retweet by pedestrian safety groups and other groups who never posted anything about Tesla. So the reach of that video clip was wide!

Think about the Waymo video of the cone debacle. How the video got hundreds of thousands of views on YT and spun hundreds of articles spanning millions of reads. This isn't something far fetched as Gali is the one saying it. Plus we know they are in constant communicate with the Beta Tester and constantly email and respond to them and have called each and every beta tester.
 
  • Disagree
Reactions: 82bert
#FSDBeta 10.0.1 St Vincents Round Trip

Looks like more Bay Area overfitting for the neural network to predict this lane is a left turn lane, but it seems like Beta 10.0.x is more aggressive in changing out of it to a straight lane even if it needs to switch lanes in the middle of the intersection. The path prediction did briefly show a left turn but it was also in the process of switching right, so it avoided a disengagement.
avoid left.jpg


Notice how the solid white line is visualized just ahead of the vehicle while directly adjacent was still dashed. Typically in California, solid white lines before an intersection separate turn lanes from straight lanes, so FSD Beta believes it's in an outer double left turn lane. This seems to also show how quickly lane change decisions are made -- basically immediately once perception determines the lane properties.