Local Image to 3D (<2gb RAM, <20s, Apple Silicon, iPhone)

My swift-mlx/python mlx port of Hunyuan3D-Paint and Hunyuan3D-Shape is finally complete! Some quick benchmarks in FP16 on my M4 Max: run wall time peak memory hy3d shape (small) 20.9 s ~5.6 gb hy3d shape (large) 22.3 s ~7.3 gb hy3d paint (rgb) 231 s ~38 gb hy3d paint (pbr) 344 s ~39 gb This (MLX) makes it possible to run the model on all recent Macs and even iPhones in Q4 or Q8, and more efficiently w/o the overhead of pytorch or even worse, CPU. What you would do with this? I honestly don't really know, ma

添加评论
点赞收藏
点踩分享查看原文
评论
?
参与讨论