Pixal3D

Pixal3D

ComfyUI-ready files for Pixal3D, packaged by Comfy-Org, with workflows that go from a single image or several views to a 3D model. Built on the TencentARC Pixal3D model.

Made by Comfy-Org on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This repository holds ComfyUI-ready copies of the Pixal3D model files, repackaged by Comfy-Org from the TencentARC original. It has BF16 and INT8 versions of the model, a multiview version, a vision encoder and two TRELLIS.2 VAE files. Two ready-made ComfyUI workflows are listed, one from an image and one from several views.

Who it suits: ComfyUI users and 3D creators who want to try image-to-3D generation inside a node editor they already know.

Industry | Who uses it | How they use it | Example
Games | Indie developers | Turn concept images into 3D assets for a prototype | A prop made from one reference picture
3D art | Artists | Start a model from images before cleaning it by hand | A character base mesh from several views
Product design | Designers | Preview an object in 3D from photos | A mug seen from multiple angles
Education | Teachers and students | Explore how image-to-3D pipelines work in ComfyUI | A classroom demo of a workflow

If you do not use ComfyUI, you can skip it, since these files are packaged for that tool.

Where to use it:

• ComfyUI on Windows, macOS or Linux with Python
• The two Comfy-Org workflow templates, named Pixal3D and TRELLIS.2 Image to Model, and Pixal3D Multi Views to 3D

Use case | Who | What they get back
Image to model | 3D creators | A 3D result generated from one image
Multi views to 3D | 3D creators | A 3D result built from several views of an object

How to set it up:

• ComfyUI installed on your computer
• Python, to run the Hugging Face download tool
• Room for the model files you choose

1. Install the Hugging Face hub tools with pip.
2. Download the repository with the hf command.
3. Put the diffusion model files in ComfyUI's diffusion_models folder.
4. Put the vision encoder in clip_vision and the two VAE files in vae.
5. Open one of the two workflows in ComfyUI.

pip install -U huggingface_hub
hf download Comfy-Org/Pixal3D

How to use it: Open ComfyUI, load one of the two workflows, point its loader nodes at the files you placed, and run it. Use the first workflow for a single image and the second for several views.

Tool | What it does
Pixal3D and TRELLIS.2 Image to Model | Workflow that builds a 3D model from an image
Pixal3D Multi Views to 3D | Workflow that builds a 3D model from several views
pixal3d_bf16.safetensors | The model in BF16
pixal3d_int8_convrot.safetensors | The INT8 ConvRot version of the model

Example: You load the Image to Model workflow, pick a picture of a single object and run it. By the workflow's name, ComfyUI returns a 3D model of that object.

For newcomers and experts: Start with the single-image workflow and one clear picture of an object on a plain background. Then try the multiview files once the first run works.

The repository splits the model into parts that ComfyUI loads separately. Diffusion models, a DINOv3 vision encoder and two TRELLIS.2 VAEs for shape and texture each go in their own folder. The INT8 ConvRot files are the lighter alternative to BF16.

Place the files in ComfyUI folders
Open a Pixal3D workflow
Load an image or several views
Run the nodes
A 3D model comes out

Good to know:

• License: MIT
• Cost to run: none named
• Data it reaches: downloads files from Hugging Face, then runs on your own machine in ComfyUI
• Not for: use outside ComfyUI, since the card describes only ComfyUI file placement

On Hugging Face, 13 November 2026

• Library: diffusion-single-file
• Downloads in the last 30 days: 856,018
• Likes: 133
• License: mit
• Based on: TencentARC/Pixal3D
• Tags: diffusion-single-file, comfyui

Links

• Model: https://huggingface.co/Comfy-Org/Pixal3D
• Files: https://huggingface.co/Comfy-Org/Pixal3D/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download Comfy-Org/Pixal3D

Then follow the usage in its model card. Nothing in this kit runs by itself.