The on-screen driving visualization is one of Tesla’s iconic features, providing drivers with a real-time view of how their vehicle perceives the world around them. For years, however, that view has been populated with generic models - all buses look the same, and some complex objects, like trains, appear as glitchy lines of semi-trucks moving at high speed.
For the first time, that’s finally about to change.
Tesla hacker greentheonly has decompiled software update 2025.38 and uncovered a massive library of new, highly specific 3D assets that are expected to be used in future visualizations. Although they’re not active yet, their presence suggests that Tesla is about to overhaul FSD visualizations.
A More Detailed World on Wheels
From its earliest days, the FSD visualization has been able to identify a surprisingly large number of objects, from different types of vehicles (sedans, trucks, motorcycles) to vulnerable road users like pedestrians and cyclists.
However, the new 3D assets cover a much wider range of objects, replacing many of the old generic models with detailed and instantly recognizable counterparts. The list of new models is extensive, so we’ve grouped them into several categories.
Emergency and Service Vehicles
FSD will soon be able to display distinct models for ambulances, fire trucks, garbage trucks, and, interestingly, street sweepers. This is a massive improvement for visualizations, which have previously just rendered the majority of these vehicles as generic vans or trucks.
Identifying and acknowledging emergency and service vehicles is important to be able to either yield for them or, in the case of a street sweeper, prepare to bypass them.
Public and Commercial Transport
A dedicated train model will finally fix the conga line of semis displayed on the visualization today. This comes alongside 3D models for trams, school buses, and a European-style semi truck.
Trailers
FSD will also finally be able to render trailers. The first trailer model included is an open-top style sideboard trailer, and we expect Tesla to introduce other trailer variants, too. It is likely that with trailer mode, this is what will display behind your vehicle in the visualization.
Vulnerable Road Users
Tesla actually has a dedicated neural network to search for Vulnerable Road Users. It seems this particular network is getting an overhaul and will have specific models for scooters, strollers, skateboards, and wheelchairs in addition to what it already shows today.
This will likely enable specific treatment for specific VRUs, as treating a scooter the same as a stroller doesn’t really make sense, as one is far faster than the other.
As Green points out, if some of the images include a blur, that means that the 3D model is animated, so in this wheelchair example, you’ll actually see the user moving the wheelchair, as seen in the video below:
Another animation of Tesla's upcoming visualizations, 15 new visualizations in all:https://t.co/rseslhOktT pic.twitter.com/DEpomx8kPG
— Not a Tesla App (@NotATeslaApp) October 17, 2025
Other Small Vehicles
Tesla has also included assets for a golf cart, as well as for three-wheelers/motorized trikes. It’s likely that we’ll also eventually see models for 4-wheelers and UTVs, as well as other smaller vehicles.
The FSD V14 Connection
The timing of this discovery is no coincidence. The question isn’t just what was added, but why Tesla is adding this level of granular detail now. The most likely answer is the massive architectural leap of FSD V14.
The previous, smaller neural networks were adept at broad classifications like car, van, truck, pedestrian, or cyclist. However, the new 10x parameter FSD V14 model is vastly more powerful, enabling the vision system to move beyond these general categories into highly specific and nuanced sub-classifications. The AI can now learn the distinct visual features that differentiate a fire truck from a generic truck, or a stroller from a shopping cart.
Requirements
It’s not clear whether these new visualizations will be available on HW3 and Intel infotainment units. While the models are fairly simple, they may require FSD V14 to recognize and display these models.
However, with the theory that Tesla is close to merging FSD and Autopilot, where Autopilot essentially becomes a limited version of FSD, they could make their way to HW3 vehicles as well.
The Curious Case of Missing Cones
The visualization's evolution hasn't always been a straight line. With the major shift from FSD V11 to the end-to-end neural networks of V12, users noticed that several common objects, most notably traffic cones, garbage cans, and speed bumps, vanished from the screen (all current FSD visualizations). They remained conspicuously absent in V13 and so far in V14 as well.
This disappearance demonstrates how deeply the visualizations are tied to the networks. The V12 rewrite required the system to re-learn how to identify these objects. In a promising sign, traffic cones and garbage bins recently reappeared in a China-specific software update, suggesting Tesla is validating their return.
These new 3D models are the necessary visual output for this newfound, deeper understanding. They help build driver trust and serve as the first tangible evidence of FSD V14’s improved perception capabilities in action.
Seeing your car correctly identify a stroller on the sidewalk or a fire truck approaching from behind does more than just look impressive. It builds trust in the system's capabilities and provides reassurance that the AI is seeing the world with the same, if not greater, level of detail as the human driver.
