1
Describe or provide a reference
Start with a prompt that names the object, style, proportions, and useful details, or upload an image that communicates the silhouette and visual direction. Meshy AI interprets this input as a design brief rather than a finished specification.
2
Generate a first 3D result
The generator estimates form, topology, camera-facing details, and materials to produce an initial model. A short prompt usually creates a broad concept, while more specific wording gives the result stronger constraints.
3
Inspect, refine, and export
Review the shape from multiple angles, adjust the prompt or texture direction, and choose the version that best fits the project. Meshy AI is most useful when generation is treated as an iterative starting point instead of a one-click final asset.