Photo3D is a browser-based image-to-3D model platform whose core function is to convert a single photo into a textured GLB 3D model. Users can start a generation task by uploading a local image or pasting a public image URL. The platform provides a free public queue that can be tried without registration, but the public GPU may sleep, be rate-limited, or have queued waits.
In terms of main capabilities, Photo3D provides more than 40 image-to-3D model channels, covering models such as Trellis, Hunyuan 3D, Tripo3D, Rodin, and Meshy, allowing users to choose different channels for reconstruction and compare results. Some channels support multi-view reference slots, allowing front, back, left, and right reference images to be added to reduce guessing about hidden surfaces and thin parts. Each model channel lists the required credits in advance, ranging from the 70-credit TripoSR draft to the 800-credit Meshy 6 channel. After generation is complete, a downloadable textured GLB file is returned, and tasks are saved in account history for side-by-side comparison.
In addition, Photo3D provides more than 50 free browser-based tools, including a 3D model viewer (supporting local viewing of GLB, OBJ, STL, and FBX), a 3D viewer and editor (which can arrange models, transform selections, and export GLB, glTF, OBJ, PLY, and STL), a GLB structure inspector, a 3D format converter (supporting local conversion of 3MF, GLB, glTF, OBJ, PLY, and STL), a texture inspector (verifying texture dimensions and GPU memory), a real-world scale calculator, an STL heightmap generator, image-to-3D model, PNG-to-3D model, and more. These local tools run in the browser, do not upload files to a server, and do not consume generation credits.
In terms of usage, users first upload or paste a source image, then choose a 3D model channel, and after generating a 3D draft, inspect the geometry, texture recovery, and output readiness, and finally open or download the GLB model in the workspace or account activity. The platform recommends source images with a centered object, a simple background, and a complete outline so that the image-to-3D model can better infer shapes. Users can also use pixel-aligned channels to ensure source fidelity, use a second reconstruction for comparison, or use multi-view channels when front, side, and back references are available.
Applicable scenarios include generating textured 3D drafts from a single product photo, object image, or concept reference for use in viewers, editors, converters, or product pages. The platform supports reading GLB, glTF, OBJ, STL, PLY, FBX, 3MF, PNG, JPG, and WebP formats, making it easy to verify texture memory and real-world scale before paid runs.
In terms of limitations, local free tools and browser previews require no account, but saving generation history, account credits, or managing AI 3D tasks requires login. The public queue requires no login or credits, but the public GPU may sleep, be rate-limited, or queue. Commercial use of generated 3D assets depends on the model channel that produced the file, the provider terms behind that channel, and the rights the user holds to the source photos. Photo3D does not claim ownership of uploaded content, but the relevant terms should be checked before using GLB in products, games, advertising, or client deliverables. Failed tasks can request a credit refund according to the refund policy by emailing support@photo3d.org with the account address and the task ID from the activity list.
