OPEN LINK ↗
// 1h agoOPENSOURCE RELEASE
Hi3D 1.0 Opens High-Res 3D Generation
Hi3D 1.0, formerly Sparc3D, is now available with inference code and model weights for high-resolution, image-conditioned 3D geometry refinement. The release supports up to 1536³ sparse extraction and was trained on approximately two million curated 3D assets. [Hugging Face](https://huggingface.co/hi3d-official/Hi3Dv1.0)
// ANALYSIS
Hi3D 1.0 makes serious local 3D generation more accessible, but its “consumer GPU” positioning needs qualification: the documented validation used a 96GB NVIDIA H20, and the release is a refinement pipeline requiring a reference image plus coarse mesh.
- –Ships source, weights, Docker configuration, and documentation through GitHub and Hugging Face.
- –Claims strong multi-view fidelity and a 90% slicing success rate, though independent benchmarks and reproduction tooling are limited.
- –Requires Linux, Python 3.10, CUDA, and an NVIDIA GPU with compute capability 8.0 or newer; there is no CPU inference path.
- –The custom Hi3D Open License permits commercial use but requires attribution and additional licensing above specified usage or revenue thresholds.
- –Its open release could accelerate 3D-printing, game-development, VR/AR, and industrial-design workflows without sending assets to a hosted API.
// TAGS
hi3d-1-03d-genopen-weightsopen-sourceself-hostedlocal-firstvision
DISCOVERED
1h ago
2026-10-25
PUBLISHED
1h ago
2026-10-25
RELEVANCE
8/ 10
AUTHOR
AI Search