Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
This site may earn commission on affiliate links.
I agree. The detection distance almost doubled -- not great but definitely improved.

Note that when AP2 first shipped back in like the 17.4 ish days, the ultrasonic range was also fairly long, but it would frequently false positive on rough road surfaces. I had a feeling they dialed back sensitivity because of that.

These days, though, it's both long range and also pretty rarely falses.

AP2 first shipped long before 17.17.4, but that was the last stable functional build until Karpathy was brough on to fix the long delayed mess that is AP2.

If I recall correctly, ultrasonics first appeared in (8.0) 2.48.204, but it might've actually been activated slightly before that firmware as my car was delivered with ultrasonics after Highland Park loaded that update (my car was, if I recall, on a week 44 build but I never saw it). AP2 only arrived (for the non-1000), on (8.0) 2.52.120 and it was heavily geofenced and limited to 35mph (which meant only stop and go and it was as useless as the Audi L3).

The ultrasonics were very limited and people measured them as the same range as AP1 and there was a big discussion over whether Tesla just installed the AP1 sensors and said AP2 could do better but now I am convinced. I had the front parking sensor issue and it was resolved and now the sensors can see much further AND the IC reflects that information (whereas I believe the sensors worked but the IC was not configured to display that data so perhaps it wasn't also shared with the APE who knows?).

TLDR: Yay sonar.
 
@croman Does that mean that in the IC the colored warning signs are now showing much earlier than previously ?
For objects further away ?
(just an oldfashioned AP1 user over here)
Well, the white outline phase shows up maybe 3-4ft earlier than with AP1.

Noticeable improvement since 10.4. Before 10.4, it was hard to see the difference.
 
  • Like
Reactions: robertvg and croman
Well, the white outline phase shows up maybe 3-4ft earlier than with AP1.

Noticeable improvement since 10.4. Before 10.4, it was hard to see the difference.

That also matches my experience. I think it sees even further with ultrasonics based on lane changes now. It's just so much better changing into tight areas and smoothly tracking the adjacent closest car.
 
Interesting post on the private UK owners FB group yesterday, where a guy stated he put is MX into reverse and got the view from the drivers-side front indicator cam instead of the BUC. I think he mentioned he was on .12

Anyways, he rebooted his MCU and the BUC returned. Unfortunately he didn't think to take pics, but he speculated the bug might be something to do with the "360" view mode that's supposed to be incoming.
 
Interesting post on the private UK owners FB group yesterday, where a guy stated he put is MX into reverse and got the view from the drivers-side front indicator cam instead of the BUC. I think he mentioned he was on .12

Anyways, he rebooted his MCU and the BUC returned. Unfortunately he didn't think to take pics, but he speculated the bug might be something to do with the "360" view mode that's supposed to be incoming.
Did he indicate if he was MCU 1 or 2?
 
Right, but if this happened on MCU1, it would be an indication that the 360 view was also coming to MCU1 as opposed to just MCU2 as I assume the compositing of the images would occur on the MCU and not the AP Hardware

How would the MCU do that; it doesn't have any direct access to the cameras. And I can't see the APE streaming the camera images over the car network, it would make the most sense for it to compost them and deliver the feed over the existing camera feed to the MCU.
 
How would the MCU do that; it doesn't have any direct access to the cameras. And I can't see the APE streaming the camera images over the car network, it would make the most sense for it to compost them and deliver the feed over the existing camera feed to the MCU.
I don't know how the connection between the APE and MCU works, but now that I think about it what you said makes more sense. I wonder if there will be a difference between AP2 and AP2.5 images since AP2.5 is supposedly full color and uses different cameras
 
Do HW2 cameras have superior dynamic range to the human eye?

Apparently, some HDR video cameras do. How do the specs of the HW2 cameras compare? Is it really fair to say that some HDR cameras surpass the human eye?

I doubt it. AP1 used the camera in monochrome mode, and at fairly low resolution since that’s all the hardware could process.
 
I doubt it. AP1 used the camera in monochrome mode, and at fairly low resolution since that’s all the hardware could process.

You mean the computing hardware? HW3 will have 400x as much computing power as HW1.

Regardless of the computing hardware, I’m curious to know what the cameras themselves are capable of.

Edit: these low-light photos by stopcrazypp from earlier in the thread look pretty good
 
Last edited:
the dynamic range of the hw2 cameras is pretty good as was already established. it might not compare all that well with full range of human vision, but human vision is kind of laggy in the sense of after looking at a bight light it is kind of blind in the dark for quite some time, the cameras do not have this problem. But a human can see a lot more in complete darkness after some time.
 
I doubt it. AP1 used the camera in monochrome mode, and at fairly low resolution since that’s all the hardware could process.

Neither resolution nor monochrome mode have anything to do with dynamic range, which is the difference in intensity between the brightest bright light that doesn't wash out and the darkest level that isn't seen as pure black.

I don't think the AP cameras are better than human eyes overall - but I don't think there's much question that they have substantially higher dynamic range from the output we've seen.
 
Really! What convinces you of this? Should I look through this thread for more images?

I'm not sure they are in this thread, but the Tesla dashcam footage driving into the sun is most impressive, and the nighttime footage is pretty revealing, too. The car sees the darker areas fairly clearly in the face of much brighter areas, and doesn't seem to blow out most of the bright parts either.