Hunyuan3D workflows: image-to-3D and API options in ComfyUI

This overview introduces Hunyuan3D workflows in ComfyUI for creating models to use in Blender. The linked image-to-3D workflow identifies version 2.1. It covers image-to-3D generation, text prompts, topology reduction and splitting a model into separate components.

The speaker describes image-to-3D as an open source process with separate geometry and texture stages, and mentions a Blender plugin. The supplied workflow description recommends two input images for basic models and four for better accuracy. For viewers planning to run models locally, the speaker cites 5 GB of VRAM for a lightweight option and 6 GB for standard shape generation.

The text-to-3D workflow uses a prompt to describe the model. The description lists versions 3.1 and 3.0 and says the newer version improves character detail. Smart topology reduces model complexity with triangles for rendering and storage, while decomposition separates a model into parts while retaining its initial geometry, according to the description.

The supplied links identify the newer text, topology and decomposition options as API workflows, distinct from the linked 2.1 image-to-3D workflow. The tutorial does not establish that every workflow runs entirely on the viewer's hardware or works offline.