3D reconstruction
Also written: 3D model generation from images
Creating 3D models from 2D images, used by Meta AI's SAM 3D for objects and humans.
3D reconstruction is the process of turning flat, two-dimensional images into three-dimensional models. Think of it like taking a photograph and using it to build a digital sculpture of the object in the picture.
This technology can be used for many things, like creating virtual reality environments, designing products, or even helping robots understand the physical world around them. Meta AI's SAM 3D, for example, can create detailed 3D models of objects and people from just a single photo.
3D reconstruction is the process of inferring the three-dimensional structure of a scene or object from two-dimensional data, typically images. This can involve techniques like photogrammetry, structure from motion (SfM), or deep learning-based approaches that learn to predict depth or full 3D representations from single or multiple views.
Recent advancements, such as Meta AI's SAM 3D, demonstrate the ability to generate high-fidelity 3D models from single images, overcoming previous limitations that often required multiple viewpoints or specialized hardware. The trade-offs involve accuracy, computational cost, and the ability to handle complex geometries, textures, and lighting conditions.
Meta AI recently announced SAM 3D, a new model that excels at 3D reconstruction from single images. This highlights a significant leap in AI's ability to understand and represent the 3D world from limited visual input, pushing the boundaries of what's possible with computer vision.
- SAM 3D: Behind the two-model design | AI at MetaMeta · AI at Meta on YouTube ·
- Introducing SAM 3D: a New Standard for 3D Object & Human Reconstruction from a Single ImageMeta · AI at Meta on YouTube ·
- Meta Meta AI
- Meta AI has developed SAM 3D, a model specifically designed for 3D object and human reconstruction from single images.
- 3D Scanning
- This typically involves specialized hardware like lidar or structur
- 3D Modeling
- This is the broader process of creating 3D digital assets, which ca
The concept of 3D reconstruction has roots in computer vision and computer graphics dating back decades, with significant advancements driven by research in photogrammetry and structure from motion. Recent deep learning approaches have further accelerated progress. Source ↗
While direct 3D reconstruction from a single image might not be a common feature in current public AI chatbots, you can explore related concepts. Try asking an AI like ChatGPT or Gemini to describe an object in a photo in extreme detail, focusing on its shape and form. You can also look for apps or online tools that offer 'photo to 3D' conversion to see basic examples of this technology.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- PromptRole-play a student struggling with math for AI tutor simulationSimulate a student's math struggles and uncertainty for AI tutor evaluation. · 5 min
- TemplateReusable workflow templates for healthcare teamsTeams can create and share templates for common workflows like documentation or prior authorizations. · 15 min
- TemplateTemplate for building an AI tool tailored to your workA structured approach to defining and building a custom AI tool for specific business needs. · 15 min
- TemplateEvals CSV prompt list for agent skillsA CSV file template for listing prompts used in evaluating agent skills. · 20 min
- HeuristicDouble coding agent task completion with docs linksProviding coding agents with a documentation link significantly increases their task completion rate. · 2 min
Questions people ask
- Can AI create a 3D model from any photo?
- Current AI models are getting better, but success depends on the photo's quality, the object's complexity, and the AI's training data. Some models, like Meta AI's SAM 3D, show promising results from single images.
- What's the difference between 3D reconstruction and 3D modeling?
- 3D reconstruction uses existing 2D data (like photos) to automatically generate a 3D model. 3D modeling is the broader term for creating 3D models, which can be done manually by artists or through reconstruction.
First seen on Subvolts Nov 19, 2025. Updated Sep 15, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.