Meshy 7

Meshy 7 is an image-to-3D generative model developed by Meshy and released as the successor to Meshy 6. It converts a reference image into a textured 3D asset, with geometry alignment—the correspondence between the generated shape and the source image—as a central design objective.

Geometry Alignment

Meshy reports three main changes aimed at improving alignment: an image encoder that processes features at multiple scales and accepts higher-resolution inputs, training data rebuilt around more precise correspondence between images and target geometry, and the use of geometry alignment as an evaluation signal during training. Its geometry benchmark evaluates global proportions, the spatial distribution of individual parts, and fine surface details by comparing generated assets with held-out reference geometry.

Mesh and Material Controls

The model supports standard and low-poly generation, triangle or quad topology, configurable polygon counts, automatic symmetry handling, and optional remeshing. Texturing can include base-color textures and PBR material maps such as metallic, roughness, and normal maps. Character-oriented controls include A-pose and T-pose generation, while rigging, animation, and higher-detail Ultra Mode are optional capabilities.