Tesla has shown off a new trick in its Full Self-Driving software: the ability to see through lighting conditions that would leave a human driver completely blind.
Elon Musk posted side-by-side images on May 9 showing what the difference looks like in practice. One image was a washed-out mess of sun glare. The other, processed by Tesla’s AI, was sharp and clear, with cars and lane markings visible despite the same blinding light hitting the camera.
The way it works is that instead of processing camera footage the way a standard image pipeline would, Tesla’s neural networks are fed raw photon counts directly. A traditional camera converts incoming light into a colour image meant for human eyes, and in the process it loses detail in the bright spots, turning them into solid white. By skipping that step and working with the raw light data before it becomes a picture, the AI can reconstruct what’s actually there, even when it looks like pure glare to anyone watching.
Musk pointed to this as a key reason FSD is getting better at handling tricky conditions like direct sunlight and nighttime driving.
It is also a pointed reminder of why Tesla has stuck with a cameras-only approach while rivals have leaned on lidar and radar to compensate for what cameras struggle with. The argument Tesla is making is that with enough AI and the right data, a camera can actually see better than a person, not just differently.
While this capability is a technical milestone, some owners on Hardware 3 vehicles note that FSD still encounters challenges when the sun shines directly into the lens, suggesting the hardware generation may still play a role in how this data is handled.
Want to see more of our stories on Google?
P.S. — Buying a new Tesla? Click here to save $1,000 USD, while supporting independent news.
Help support us by shopping on Amazon here.
Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent media!