weiyaow

Fun visually-grounded 3d visualization with Muse Spark - by asking the model to generate a 3d visualization of how much each component cost on a xbox 360 motherboard (and put the bar on the component).

Excited that this is out, and proud to be a part of it! This is our first step in scaling — an intelligent and useful multimodal model 🥑 built on a foundation of scientific rigor and solid engineering. Incredible velocity and culture of the team and the best is yet to come!

Huge thanks again to Weiyao @weiyaow1 and Xitong @XitongYang1 for their incredible talks! 🚀 The recordings are officially LIVE. Dive into the insights and catch up on what you missed here👇 📹 Watch now: youtu.be/D4MOJx9om7I

🎙️The AI Talks | S5E6 Introducing SAM 3D: Powerful 3D Reconstruction for Physical World Images — by Weiyao Wang @weiyaow1 & Xitong Yang. Can we "3D-fy" everything? 🌍🏃‍♂️ From universal objects to human bodies, explore the new SAM 3D family. Join us👇

My team at MSL is hiring interns to work on World Models! 🌎 We recently released SAM 3D ai.meta.com/sam3d/ Apply & DM me metacareers.com/profile/job_de…

My team is hiring Research Scientist Interns this year! The team is part of the Meta Superintelligence Lab (MSL) and has launched SAM 3D in 2025. Come to join us for our next big bet! Reachout to us with a CV or apply directly through the job posting! Job post: metacareers.com/profile/job_de… SAM 3D: ai.meta.com/sam3d/

We’re in San Diego this week for #NeurIPS2025! Stop by the Meta booth (#1223) to meet our team and check out: 🔎 Demos of our latest research including DINOv3 and UMA ⚡ Lightning talks from researchers behind SAM 3, Omnilingual ASR and more (see schedule below) 👓 Hands-on demos with our latest AI glasses including the Meta Ray-Ban Display Our team is also sharing 19+ papers and 13+ workshops this week. We hope to see you there.

@loocaslu We currently do not have plans to support API; we may be able to visit this option. If there are specific issues encountered, feel free to create an issue on GitHub and we can help take a look. Thanks again for trying out our work!

@weiyaow1 any plans to deploy an API access for SAM 3D? we downloaded the open source model but haven't been able to spin up a containerized instance in Replicate for 2D to 3D inference

Meta released SAM3 a model to reconstruct objects from segmented images. I tried it and here are 4 things I found out - The generation is incredibly fast when compared with other 3D GenAI models - The models generated are not meshes but Gaussian Splatting. I was surprised that in the paper it was not mentioned anywhere - When generating multiple objects the arrangement is a bit messed up. - Unfortunately, it is impossible to download any of the .ply generated 😔 I cannot stop thinking about the implications of this in XR. Objects in the real world could be almost instantly turned into photorealistic copies and even rearranged to the user's wish.

so cool!

Today we’re excited to unveil a new generation of Segment Anything Models: 1️⃣ SAM 3 enables detecting, segmenting and tracking of objects across images and videos, now with short text phrases and exemplar prompts. 🔗 Learn more about SAM 3: go.meta.me/591040 2️⃣ SAM 3D brings the model collection into the 3rd dimension to enable precise reconstruction of 3D objects and people from a single 2D image. 🔗 Learn more about SAM 3D: go.meta.me/305985 These models offer innovative capabilities and unique tools for developers and researchers to create, experiment and uplevel media workflows.

The Segment Anything Playground is a new way to interact with media. Experiment with Meta’s most advanced segmentation models, including SAM 3 + SAM 3D, and discover how these capabilities can transform your creative projects and technical workflows. Check out some inspo and tips in the 🧵 below, then head over to the Playground to get started: aidemos.meta.com/segment-anythi…

SAM 3D enables accurate 3D reconstruction from a single image, supporting real-world applications in editing, robotics, and interactive scene generation. Matt, a SAM 3D researcher, explains how the two-model design makes this possible for both people and complex environments. 🔗 Read the SAM 3D Objects research paper: go.meta.me/8c08ca 🔗 Read the SAM 3D Body research paper: go.meta.me/5e60ed

昨天 Meta 开源一个 SAM 3D 模型，完全被 Gemini 3 Pro 的消息淹没，但我觉得非常值得关注一下。 平时大家看到喜欢的物品想做成 3D 模型，传统建模软件学习成本高、操作复杂，拍多角度照片还很麻烦。 而 Meta 此次开源的 SAM 3D Objects 项目，只需一张照片就能重建出完整的 3D 模型，包括形状、纹理和空间位置。 尤其擅长处理真实场景中的复杂情况，即使物体有遮挡、尺寸很小，或者场景很杂乱也能准确重建，还支持同时重建多个物体并保持它们的空间关系。 GitHub：github.com/facebookresear… 此外，项目还提供了在线演示和完整的 Jupyter 示例代码，可以导出 3D Gaussian Splatting（高斯泼溅）格式方便后续处理。 按照文档配置环境后即可使用，非常适合电商产品展示、室内设计、游戏场景制作等需要快速 3D 建模的场景。

@loocaslu Yes they are the same. In the demo we use a lighter version of the Gaussian Splatting Decoder, to give a smaller amount of the splats to be browser friendly. The OSS has the one that gives the full Gaussian Splatting predictions.

@weiyaow1 amazing work! is the open source release the same model as the one on segment anything playground demo? aidemos.meta.com/segment-anythi…

SAM3のCreate 3D Scene機能、写真からオブジェクトを1個１個選んで3D作成を押していくだけで街ができていく。この方向で進化したらドチャクソ便利やん。その小さな物体、車って分かるの！？って感じだった。

中々の精度です！ Metaが画像から瞬時にオブジェクトの3D化や人体形状の推定を可能とするAI技術を発表！ ブラウザ上でお試し可能！ SAM 3D 3dnchu.com/archives/sam-3… #SAM3D

oh my.. hows this even possible Meta just dropped SAM 3D, you can auto select any object in still image and.. turn them into high quality 3D model the model is free to download and test in playground link in comments

3Dfy anything from a single image! Very thrilled to announce SAM 3D. From an input image, select any object you want, 3Dfy it! Blog: ai.meta.com/blog/sam-3d/ Demo: aidemos.meta.com/segment-anythi…

The implications here for robotics data generation is huge.