# SAM 3D Objects Canonical page: https://fuser.studio/models/sam-3d-objects SAM 3D Objects in Fuser turns an object in a photo, picked out with a text prompt, into a textured 3D mesh. Refine and rig it with the other 3D nodes. ## Specs | Spec | Value | | --- | --- | | Inputs | Image, Prompt (text) | | Output | 3D model | | Cost | 28 credits per run | Turn casual single-view photos and cluttered scenes into production-ready 3D assets with prompt-driven segmentation and baked UV textures Creator: Meta AI [Generate 3D asset](https://app.fuser.studio) ## How SAM 3D Objects works Upload a photo, pinpoint the subject with a short noun prompt, and export a textured GLB mesh in seconds. ### Upload your source image Provide a high-resolution reference photo of the physical product or scene taken under diffused, even illumination. ### Target the item Enter a short noun phrase to isolate your subject and fine-tune the detection confidence slider between 0.1 and 1.0. ### Download spatial GLB Export a textured GLB mesh complete with baked UV maps and synthesized back geometry for spatial apps or DCC pipelines. ## What SAM 3D Objects is good at From prompt-guided segmentation to automated volumetric completion, turn ordinary reference images into production-ready spatial assets. ### Cluttered-scene object isolation Isolate specific items in dense, cluttered living spaces or tabletop scenes using targeted noun prompts and adjustable detection thresholds spanning 0.1 to 1.0. ### Single-view geometry completion Synthesizes coherent, solid back faces, bases, and hidden geometries for objects viewed from only a single camera perspective using diffusion-based spatial priors. ### Baked UV texture export Generate unified GLB meshes with baked UV coordinates and surface texture projection ready for immediate import into WebGL, Unity, or Unreal Engine. ### Instant product digitization Extract and digitize real-world furnishings, decor, and manufactured goods directly from standard reference photos without multi-angle photogrammetry rigs. ## Made with SAM 3D Objects A gallery of textured 3D meshes reconstructed from single-angle source imagery, spanning furniture, industrial prototypes, and sculpted decor. ### Industrial seating prototype with baked wool textures ### Artisan tabletop decor with complete rear geometry ### Sportswear product model with baked material maps ### Spatial decor asset isolated from surrounding clutter ### Heritage archival artifact with organic surface detail ## What people build with SAM 3D Objects Designed for 3D generalists, spatial developers, and e-commerce teams converting real-world inventory into lightweight virtual assets. ### E-commerce product digitization Convert studio photos of catalog inventory into interactive 3D web viewers and AR preview models without manual retopology. ### Game asset prototyping Transform single concept illustrations or real-world prop photos into spatial blockout meshes for rapid level dressing. ### Interior scene staging Extract architectural furniture, designer lighting, and interior accessories directly from scene photographs for floorplan populating. ### Footwear and apparel design Turn footwear and accessory mockups into textured 3D models for virtual fitting rooms and digital lookbooks. ### Spatial web experiences Generate lightweight textured assets from reference images for immersive Three.js, WebGL, and Apple Vision Pro experiences. ## Frequently asked questions ### How does SAM 3D Objects convert a single photo into a 3D model? The model takes a single 2D photograph alongside an optional text label, segments the target item using Segment Anything segmentation, and reconstructs a complete 3D mesh (GLB) with volumetric back-surface hallucination and baked UV textures. ### What prompt format works best for object segmentation? Keep text prompts to tight noun phrases naming the exact item (such as "leather armchair" or "ceramic vase") rather than descriptive narrative sentences. The text prompt is used by the segmentation engine to pinpoint the item in crowded photographs. ### When should I enable Export Textured GLB? Toggle Export Textured GLB on when building for WebGL, Three.js, game engines, or AR viewers that require standard baked UV coordinates and surface texture maps. Turn it off when you only need raw untextured spatial geometry to remesh or retexture in external DCC software. ### How do I prevent background clutter from fusing into the 3D mesh? Adjust the detection threshold slider between 0.1 and 1.0. For crowded scenes with overlapping clutter, keep the threshold high (0.6 to 0.8) to isolate the silhouette cleanly; lower it (0.2 to 0.4) only when isolating faint, low-contrast, or slender structures. ### What object types and lighting conditions should I avoid? Avoid transparent glassware, mirror-finish chrome, complex mechanical assemblies requiring CAD-level tolerances, and photos taken under harsh single-direction lighting that permanently bakes dark drop shadows into texture maps. ## Try SAM 3D Objects on Fuser Turn casual single-view photos and cluttered scenes into production-ready 3D assets with prompt-driven segmentation and baked UV textures [Generate 3D asset](https://app.fuser.studio) ## Related models - [Meshy](https://fuser.studio/models/meshy.md) - [Rodin](https://fuser.studio/models/rodin.md) - [Tripo 3D](https://fuser.studio/models/tripo-3d.md) - [Hunyuan3D V3](https://fuser.studio/models/hunyuan3d-v3.md) - [Trellis 2](https://fuser.studio/models/trellis-2.md) - [Hunyuan3D Multiview](https://fuser.studio/models/hunyuan3d-multiview.md) Node reference: [SAM 3D Objects docs](https://docs.fuser.studio/docs/nodes/mesh/sam-3d-objects.md) ## More articles - [Best AI 3D Model Generators (2026)](https://fuser.studio/articles/best-ai-3d-model-generators.md)