meshy-3d

What it does

Converts prompts or reference images into 3D assets using Meshy’s REST API and web platform. It supports text-to-3D and image-to-3D workflows, including multi-image inputs, as well as AI texturing/PBR, remesh/topology control, UV mapping, and auto-rigging/animation. It exports to GLB/FBX/OBJ/USDZ/STL (and 3MF via remesh) for Unity/Unreal/Blender, and can drive Meshy’s asynchronous job API (auth, polling, streaming, webhooks, credits, rate limits). It also covers reviewing meshes for game/film/print readiness and deciding whether Meshy is appropriate for a given task.

How it works

• Uses asynchronous jobs: POST a task to the Meshy API and receive a task id; poll, stream, or wait for a webhook until status SUCCEEDED.
• Task lifecycle: PENDING → IN_PROGRESS → SUCCEEDED (or FAILED/CANCELED); progress 0–100; queue position shown via preceding_tasks during PENDING.
• Three task-finalization methods: (1) Poll GET /openapi/v1/<endpoint>/:id until terminal, (2) Stream GET .../:id/stream for SSE events, (3) Webhook notification on completion.
• Text-to-3D uses a two-step preview (geometry) and refine (texture) workflow with mode: "preview" and mode: "refine" calls; outputs include model_urls, texture_urls, and thumbnail_url.
• Image-to-3D and multi-image-to-3D reconstruct subject(s) from input images; texture/topology params mirror text-to-3D refine; multi-image supports 1–4 images for better reconstruction.
• Topology, poly count, and remesh controls: topology, target_polycount, decimation_mode, and remesh endpoints for repairing/topology adjustments; recommended order: Remesh → Unwrap UV → AI Texturing → Rigging → Animate → export.
• Rigging/Animation: POST to /openapi/v1/rigging with textured input; outputs include rigged GLB and FBX plus basic animations; animation library available via /openapi/v1/animations with action_id.
• Export and engine-import guidance maps formats to use-case: GLB for Unity/Unreal/Three.js; FBX for rigs; USDZ for AR; STL/3MF/others for printing; ensure scale via auto_size/origin_at or height_meters.
• Texturing: enable_pbr, hd_texture (Meshy-6), remove_lighting; retexture endpoint supports style prompts and image_style_url; UVs can be preserved via enable_original_uv during retexture.
• Quality considerations and cautions: not suitable for exact CAD tolerances or guaranteed watertight manufacturing; output should be validated in downstream tools; asset retention is ~3 days on non-Enterprise plans.

When to use it

Use Meshy-3d when you need a coherent prop/character/creature from a prompt or reference image, a textured starting mesh, or bulk asset generation where throughput and cost per asset matter more than perfect topology. Do not use it for exact real-world mechanical parts, guaranteed manifold geometry, editing existing hand-authored geometry, or full-scene reconstructions. Also, avoid combining lowpoly mode with fine topology controls when image-to-3D requires detailed topology.

What it can touch

• Endpoints to generate, refine textures, remesh, retexture, rig, animate, and export (openapi/v2/text-to-3d, openapi/v1/image-to-3d, openapi/v1/multi-image-to-3d, openapi/v1/remesh, openapi/v1/retexture, openapi/v1/rigging, openapi/v1/animations).
• Authentication via Bearer token in Authorization header; API key stored securely (not hardcoded).
• Task IDs, status polling, streaming, and webhook configuration for task completion.
• Formats for export: glb, obj, fbx, stl, usdz, 3mf (plus blend via remesh).

Caveats

• API access requires a Pro tier or higher; free tier cannot create API tasks.
• Meshy guarantees are not provided for watertight/manifold results; validate outputs in downstream tools.
• Some deprecated or non-functional parameters exist (e.g., certain symmetry or pose options); use current equivalents as documented.
• Outputs and pricing can change; verify model versions, credits, and limits before quoting to users.