Mohamed El Banani

We’re finally sharing what we’ve been up to @theworldlabs! This is the first step on our journey to build fully interactive and immersive worlds that allow you to bring your creativity to life. Check out the demos, my favorites are the Kandinsky landscape and Van Gogh terrace.

We're excited to be rolling out two model updates today! Marble 1.1: Improves lighting and contrast, with a major reduction in visual artifacts. Marble 1.1-Plus: Our new model built for scale. Create larger, more complex environments than ever before.

Introducing Marble by World Labs: a foundation for a spatially intelligent future. Create your world at marble.worldlabs.ai

Today we announced RTFM (Real-Time Frame Model) — a world model that generates frames in real time from any camera viewpoint. Unlike standard video models, RTFM understands 3D geometry. You can literally move the camera through the generated world. 🎥👇

We've been exploring different ways of modeling the world at @theworldlabs. This direction combines our real-time learned renderer with posed frames as a persistent spatial memory. I am excited to see where we go next! Check out the demo here: rtfm.worldlabs.ai

Generate persistent 3D worlds from a single image, bigger and better than ever! We’re excited to share our latest results and invite you to try out our world generation model in a limited beta preview.

Experimenting with image-to-world: creating consistent 3D environments from a single photo. As someone who studied architecture, this feels like a glimpse into the future of spatial creation and I’m excited to see where World Labs takes it.

Ever wish YouTube had 3D labels? 🚀Introducing🎥DynPose-100K🎥, an Internet-scale collection of diverse videos annotated with camera pose! Applications include camera-controlled video generation🤩and learned dynamic pose estimation😯 Download: huggingface.co/datasets/nvidi…

I’m on the PhD internship market for Spr/Summer 2025! I have experience in multimodal AI (EHR, X-ray, text), explainability for image models w/ genAI, clinician-AI interaction (surveyed 700+ doctors), and tabular foundation models. Please reach out if you think there’s a fit!

Playing around with @theworldlabs' 3D generator and combining it with @runwayml vid2vid to blend myself and the background better. Although it's only a side perspective, you can tell it doesn't have that static image feel because of how the 3D environment is moving.

The future of immersive creation is here. World Labs image to 3D worlds: • Generate 3D worlds from a single image • Navigate scenes with real-time camera control • Use different effects without changing the scene I've had access and you can see some of my early results below: @theworldlabs

I threw a couple of Mac Baconai's images in the new WorldLabs model. The scene completion is pretty darn good.

Want to push the limits of where this tech can go and build the products that put it into users' hands? Join us! jobs.ashbyhq.com/worldlabs

Generative 3D worlds are amazing! Here is a generative AI video showing fine grained camera control. Camera movements are all hand places within a generated 3D world.

I had the opportunity to try out @theworldlabs' incredible 3D environment-generation model. This is still in the early stages. Check out some of the use cases I demonstrated!

We’ve been busy building an AI system to generate 3D worlds from a single image. Check out some early results on our site, where you can interact with our scenes directly in the browser! worldlabs.ai/blog 1/n

we’re hiring @theworldlabs seeking insanely great engineers + designers to work alongside our world-class research team to imagine and build entirely new apps and experiences made possible at the rapidly expanding frontier of generative AI + 3D computer vision + graphics

Summer Update: Had an incredible research internship at @theworldlabs 🌎working towards #SpatialAI 🚀🚀 Next Up: PhD @CMU_Robotics !

@tylerraye @abhinav1kumar I agree with @tylerraye. Linear probes don't seem to do well on MAE; eg, compare the linear probe vs. finetune perf. on ImageNet. Larger decoder probable helps MAE. Another difference is evaluating image-level vs. dense features where you would expect CLIP to do better as well.

@abhinav1kumar i'd guess the ordering relates to different readouts: they use a dense multi-scale decoder while we're using a lightweight classifier. CLIPs are probably better suited for our lightweight readouts then MAEs. but curious what @_mbanani thinks about this!

do large-scale vision models represent the 3D structure of objects? excited to share our benchmark: multiview object consistency in humans and image models (MOCHI) with @xkungfu @YutongBAI1002 @thomaspocon @_yonifriedman @Nancy_Kanwisher Josh Tenenbaum and Alexei Efros 1/👀

Happy to announce our company has come out of stealth! We are building spatial intelligence with some of the most talented researchers and engineers -- @ZhuLuyang, @ericryanchan, @kdexd, @_mbanani, @KyleSargentAI, @chaoyuaw, and many more. We are hiring so please reach out!