Introducing MirageLSD: The First Live-Stream Diffusion (LSD) AI Model Input any video stream, from a camera or video chat to a computer screen or game, and transform it into any world you desire, in real-time (<40ms latency). Here’s how it works (w/ demo you can use!):

[2/] Transform into a robot or a wizard, visit a tropical island or a frozen wonderland. Just tug at your shirt to change it and swap faces by waving your hands. Use brooms as lightsabers and pens as wands.

[3/] Watch your feed in an alternate dimension — if you’re watching a gym video, why not have it set in the Colosseum? Or watch snowboarding in the desert?

[4/] Because Mirage runs in real-time, you can take your favorite video games and set them in any universe of your choosing.

[5/] You can vibe-code a game in 30 minutes, and let Mirage handle the graphics. From prompt to GTA VII, before GTA VI.

[6/] You don’t need to have expensive gear to stream with movie-quality graphics. Using a broom as a fancy microphone and a box as a mixing board, you can be who you really are.

[7/] MirageLSD runs at 24 FPS, has <40ms of latency, and can generate infinite video lengths. This required breakthroughs from CUDA Megakernels to drift-resistant training in order to achieve over 100x efficiency gains and infinite generation. A link to the technical blog post can be found below.

[8/] Use the app here: (today website, iOS and Android coming next week) Read our technical blog here:

Hm I wonder if synchronizing two video streams for binocular world building is practically possible soon (and worthwhile?) You could instantly morph anything in front of your eyes into whatever you want and have it feel real

Have you tried it with a mixed reality headset?

When we said “mind blowing consumer experiences” this is what we meant. Let’s GO @DecartAI 🚀

i tried it and its absoluetly insane. holy crap EOY we might have real time veo3 quality

Congrats on the launch, this is awesome!! Mirage never dies 🫡, @_amankishore @SreeTrips

never seen a more accurate name

However weird we think things are going to get, they'll be much weirder.

@grok is this real

Mirage let's you change the world 🌎

So StreamDiffusion but closed-source… 😂

It’s extremely good but o wonder under what hardware it’s &lt;40ms

The word “hallucination” just got a whole new meaning

i’m scared

much more interesting than what zuck is working on

Lets goo

Congratulations on your launch. Just tried it and it is fun.

Very interesting!

Congratulations 👏

The reality will be shaped by our imagination

this idea has extreme potential and someone is going to do something amazing with it eventually

We just need to implement and synchronize, stabilize etc. in good enough way communication between biological neural representations in our brain's imagination and machine neural representations in reality filter glasses

Congrats on the launch!

Congrats!

Congratulations, the following two years will be epic 👌

This technology is absolutely amazing

That is the future, right in front of us.

wow this is so cool! my first 2 questions: open model at some point? how about a quest 3 app that accesses the passthrough camera (i am available for doing r&amp;d)

fun AI is hereeeee

what do the safety guidelines on this look like? when you're using a diffusion model to output highly dimensional data, how do you ensure what's being generated isn't malicious?

Insane, tried with apple vision pro, mind blowing!

Damn 😯

Would love to test this on a Quest 3

wow this is insane