Bringing Native Support for 3D Gaussian Splats into ComfyUI with TripoSplat

1️⃣ Bringing Native Support for 3D Gaussian Splats into ComfyUI with TripoSplat

Published on Jun 01 2026

The ComfyUI team has announced the day‑zero support for TripoSplat, an open‑source 3D Gaussian splat model from Tripo. With a single image, users can now generate a lightweight 3D asset that integrates seamlessly into modern 3‑D pipelines, AR/VR experiences, and interactive applications.

What the Announcement Covers

• Model: TripoSplat (MIT‑licensed weights and inference code).
• Workflow: One‑image input → 3‑D Gaussian asset.
• Target Use‑Cases: Stylized characters, props, creative designs, and 3‑D previews for concept art or video generation.
• Integration: Available in the ComfyUI Template Library; compatible with the upcoming v0.23.0 release for both cloud and desktop.

Key Features Highlighted

Why This Matters for Creators and Template Authors

• Efficiency: One image to a ready‑to‑use 3‑D asset cuts asset creation time dramatically.
• Optimisation: Artists can fine‑tune the balance between visual fidelity & file size—critical for mobile or web‑based projects.
• Workflow Expansion: Template authors can now bundle TripoSplat workflows with their existing collections, offering users a complete 2‑D‑to‑3‑D pipeline inside ComfyUI.
• Community Growth: Open‑source weights encourage experimentation and extensions—perfect for the Contrib community on ComfyUITemplates.com.

How to Get Started

1. Update ComfyUI to v0.23.0.
2. Open the Template Library and search for TripoSplat.
3. Load the TripoSplat workflow template.
4. Follow the on‑screen guide to download the model weights (🤗 VAST‑AI/TripoSplat).
5. Upload an image and run the workflow to generate your 3‑D Gaussian splat.

Model weights: 🤗 [VAST‑AI/TripoSplat]
Inference code: VAST‑AI‑Research/TripoSplat

Further Reading

• Official Blog Post: Bringing Native Support for 3D Gaussian Splats into ComfyUI with TripoSplat
• Model Repository: TripoSplat on Hugging Face

2️⃣ Wan 2.1 Video Model Native Support in ComfyUI!

Published on Feb 26 2025

The ComfyUI team now includes native support for the Wan 2.1 large‑scale video generation model. Two distinct scale options—14 B and 1.3 B parameters—are now available, with full accessibility on consumer GPUs.

Overview of the Update

Significance for the Community

• Lower Barrier to Entry: Users can now experiment with cutting‑edge video generation without needing high‑end GPU clusters.
• Workflow Flexibility: Template authors can provide both a lightweight and a high‑capability node, allowing creators to choose based on their hardware.
• Speed & Cost Efficiency: The 1.3 B version delivers fast inference with reduced render cost, ideal for rapid prototyping or mobile deployment.

Quick‑Start Instructions

1. Confirm your ComfyUI installation is up to date.
2. Add the Wan 2.1 node via the node catalog or community plugin.
3. Select the desired scale (14 B or 1.3 B) within the node settings.
4. Configure input prompts, seed, and frame count.
5. Execute the workflow to generate your video.

Further Reading

• Official Blog Post: Wan 2.1 Video Model Native Support in ComfyUI!

3️⃣ WAN 2.2 Animate & Qwen‑Image‑Edit 2509 Native Support in ComfyUI

Published on Sep 23 2025

The latest ComfyUI release brings native support for the WAN 2.2 Animate framework and the Qwen‑Image‑Edit 2509 model. These additions elevate pose reference handling and image‑editing capabilities within the platform.

Core Highlights

Why Creators Should Care

• Enhanced Animation Workflows: Animate nodes give creators precise control over pose transfer, beneficial for character rigging and motion capture pipelines.
• Image Editing Empowerment: Qwen‑Image‑Edit enables sophisticated edits (inpainting, style transfer) directly within ComfyUI, streamlining content creation.
• Template Expansion: These nodes can be incorporated into new or existing templates, broadening the appeal of collections hosted on ComfyUITemplates.com.

Getting Started

1. Update ComfyUI to the latest version that includes WAN 2.2.
2. Search the node catalog for WAN 2.2 Animate and Qwen‑Image‑Edit 2509.
3. Drag the nodes into your workflow, set up the input images and parameters, and run.

Further Reading

• Official Blog Post: WAN 2.2 Animate & Qwen‑Image‑Edit 2509 Native Support in ComfyUI

4️⃣ Introducing ComfyUI Native Partner Nodes (and New Brand!)

Published on May 6 2025

In a major expansion of its ecosystem, the ComfyUI team has launched its Native Partner Nodes platform. The initiative currently bundles 11 models across 65 nodes, offering a unified API and a fresh brand identity.

What’s Included

Impact for the Community

• One‑Stop Shop: Creators can import a wide range of models and nodes without juggling multiple plugins.
• Brand Consistency: The new brand provides a recognizable interface, simplifying discovery on directories like ComfyUITemplates.com.
• Collaboration Opportunities: Partner projects can now expose their models as native nodes, encouraging cross‑ecosystem innovation.

How to Leverage These Nodes

1. Install the ComfyUI Native Partner Node package (available via pip or the UI installer).
2. Browse the Node Library to explore the 65 nodes.
3. Load your favorite models directly into your workflows.

Further Reading

• Official Blog Post: Introducing ComfyUI Native Partner Nodes (and New Brand!)