Загрузка видео...
Не удалось загрузить видео
Thats awesome! A developer used Al-powered 4D Gaussian Splatting to convert flat video footage into dynamic 3D spatial scenes. The system reconstructs different camera angles and depth information from ordinary footage, making it possible to navigate scenes in three dimensions. We're moving from recording video to digitally recreating reality itself.
73,730 просмотров • 4 месяцев назад •via X (Twitter)
Комментарии: 26

if you wanna read: source of above video:

Wishful thinking. From the page you’re linking to, which is not project but an explanation of 4DGS. It’s not based on one 2D video stream…

finally a reason to upgrade a computer cpu with a gpu & npu. imagine watching a movie, then using such a technology to change your pov whenever you want to, while the movie continues to play at it's usual 24fps.

動画を「再利用できる空間データ」として扱えるようになるのが大きいですね。既存映像から体験型コンテンツへ変換するワークフローが生まれそう。

One step closer to the Holodeck

This is amazing lol

Gaussian splatting combined with AI can reconstruct spatial depth and multiple viewpoints from ordinary video footage. These advances in neural rendering and volumetric reconstruction are pushing digital media beyond flat recording toward immersive, navigable representations of real-world environments.

Zapruder film to 3D anyone? Latest UFO Pentagon Video?

Nonsense. Can’t be done, yet.

this.

its fake. of you want to watch imagined angles do it. i dont

source: paper?

太強了👍

How to call this? 4DGS ?

4D Gaussian Splatting sounds like sci-fi jargon but the result is pure magic. Can't wait to see this in games or VR tours.

oh hey I use a similar technique though when correct it should function with far greater detail. (based on holographics)

That’s very cool!

Now harry potter is real?

wow thats big 👏🏼👏🏼👏🏼 so you can see every angle from any movie? 🫣

2D video → 3D spatial scene. That’s not just rendering. That’s reconstructing reality. Now imagine the same leap for workflows: → Flat task description → Dynamic agent execution across tools, files, and APIs → Full spatial awareness of your digital environment That’s @EasyClaw. We’re moving from recording instructions to digitally executing work itself. Same leap. Different dimension. ⚡️

this is probably the worst version of this I've seen

Good job 👏

wow amazing

4DGS addresses the missing piece: scanning existing infrastructure into spatial formats at scale. Makes Omniverse workflows viable for real-world deployment.

I suspect the porn industry is already claiming dibs on this. 😂

>no vertical rotation ngmi
