PartCrafter vs Unique3D

PartCrafter is a free, open-source model that generates multiple 3D parts and objects from a single RGB image. It can handle individual objects and complex scenes, producing distinct meshes with meaningful part structure without requiring a pre-segmented image. Its architecture represents parts with disentangled latent tokens and uses hierarchical attention to pass information within and across parts. Local inference scripts can remove backgrounds and optionally apply style transfer to adapt real-world photos toward the rendered-image domain used for training. An optional vision-language feature can suggest the number of parts; Gemini is the default provider, and the provider architecture can be extended. Pretrained object-level and scene-level weights are hosted on Hugging Face. Installation is from the GitHub repository in a Python environment, and inference requires a CUDA-enabled GPU with at least 8 GB of VRAM. The linked Hugging Face demo is paused, though its page says users can ask the authors to restart it. The repository lists an MIT license and provides the software as-is, without warranty.

Unique3D generates textured 3D meshes from a single image. Its method derives four orthographic views from the input and progressively increases their resolution, then generates matching normal maps with a normal diffusion model. The project's ISOMER reconstruction algorithm combines color and geometric priors and can produce meshes with millions of faces. The project describes generation from a single-view image in 30 seconds and says a high-fidelity textured mesh from one orthogonal RGB image can be generated in under 30 seconds. Image-to-3D and AI texturing are listed capabilities, and GLB is the listed export format. The repository links to a ComfyUI implementation and provides instructions for an interactive local Gradio demo after model weights are downloaded. An online demo is free to try with a registration invitation code; it is described as slower but more stable than the Gradio demo. The Linux setup targets Ubuntu 22.04.4 LTS and CUDA 12.1, while Windows installation may require Visual Studio Build Tools.