oh my… this shouldn’t be possible Gemini 3 can generate 3D interactive scenes with three.js… and you can literally move particles with your hands no coding skills needed at all, it's all free tutorial + prompts in the comments

tutorial: open Google Al Studio ( and paste the prompts in next comment copy the code generated by it to Gemini 3 create a new text, copy the code, and change the file name from the original .txt to index.html. finally, just open it with a browser. Remember to turn on the camera

prompts: Create a real-time interactive 3D particle system with Three.js. requirements: 1. Control the scaling and expansion of the particle group by detecting the tension and closing of both hands through the camera. 2. Provide panels that can choose hearts/flowers/saturn/Buddha statues/fireworks and other templates 3. Support the colour selector to adjust the particle colour 4. Particles need to respond to gesture changes in real time. The interface is simple and modern. 5. The interface is simple and modern

I remember spending weeks learning GLSL shaders to get particles to move like that. Watching a prompt do it in 30 seconds is humbling.

we are witnessing the historical moment man

See what I build

Can confirm it works 👍🏾

i built the same but got much less impressions:((((

everyone who watches this is thinking this... I'm just gonna ask it.. BUT can it do THIS!!!

Return of LeapMotion! Made this video 12y ago (my first ever published Youtube video) with LeapMotion and Google Glass. LOVED IT! All that didn't really catch on. And, I got locked out of that log in so no more videos on that account. Someday I dream of getting the access back @Google please please.

Can it tell you how many particles it sees?

they said around 2b :)

Got no camera, came up with this 😊

would it work with apple vision in web xr immersive mode

yeah we can create our own apps easily for Vision Pro bro

This is incredible! The fact that Gemini 3 enables true 3D interaction with particle manipulation—no coding required—really pushes the boundaries of accessible creation tools. This feels like we're crossing into territory where the barrier between "developer" and "creator" completely dissolves. Can't wait to see what people build with this 🔥

Well this has been possible for far longer than AI. We do this in touchdesigner easily with everyday camera’s. I think just a lot of people haven’t figured out these things connect if you want them to because they are vibe prompting it. If you wanted to do anything serious though you’d graduate up to touchdesigner to do it. For web apps though it’s cool.

Ahh, so JARVIS was just the prompt. I thought Mr. Strak was a fullstack developer

That's a really informative or new thing which I will know 😅

try it and have fun mate

Can we connect

No way. What a cool idea!

possibility is unlimited bro, I’m trying to do actual 3D models with this

Can't wait to see what you create man!

It works. Anyone here interested in sharing their technology related simulation or creation or interested in building STEM modules reach out. AI will write the code, but your ideas can make a difference in STEM learning.

Bro we’re two updates away from bending reality like it’s Photoshop 🤯

Man! This is beyond amazing. Thanks for the detailed step too.

welcome mate

@ivanfioravanti Nice illusion of magic it creates.. good for artwork

Wild how Gemini 3 turns 3D scenes into no code magic

🫡🫡

amazing my friend!

Google is winning so badly bro

Yes I saw that😥That s why I try to support the platforms less well-known here such as Vidu, Pika or Pixverse. In the long term, we need them to maintain a healthy market.

Such a cool project @EHuanglu! Using your prompt, I one-shotted this with @xai grok code fast 1 model! It worked great!

This fucking sucks actually

This is a remarkable demonstration of multimodal AI integration. Gemini 3 generating fully interactive 3D scenes via three.js, with real-time hand-based particle manipulation, implies tight coupling between gesture recognition, scene graph updates, and GPU-accelerated rendering—all abstracted for no-code use. From a systems perspective, it’s a seamless pipeline combining computer vision, physics simulation, and real-time rendering, making advanced interaction accessible without traditional programming.

wow🤯, thank you very much for sharing, it really works! it is unbelievable cool to see and then to see it suddenly implemented in seconds just by prompt!

@amannhq bc this works

And I just get error. 😅

This is cool. I did the floating particles for my landing page at but the hand gesture thing is unreal.

this actually looks like an old mediapipes holistic demo from 2020?

Amazing 🌟

cool

Thank you for giving me a reason to stay on the planet. It was getting rough for a little bit there. 😭

Thank you so much! The code generated by your prompt was incomplete but @Gemini 3 helped me sort what was missing and i came up with this, not as cool as yours but still cool. Thanks bro.