Mixar is an AI-native 3D editor built on Blender that embeds agentic workflows directly into the modeling pipeline. It lets artists dispatch background agents for tasks like unwrapping, baking, and retopologizing while continuing to work, and supports layer-based texturing with editable PBR effects. The tool is open source and runs natively on Mac and Windows.
Funding
Funding not disclosed
Founders
Product
Problem
Traditional 3D modeling and texturing workflows in Blender require significant manual effort for repetitive tasks like UV unwrapping, baking, and retopologizing, which slows down creative teams and interrupts their flow. Artists often have to switch between multiple tools for reference gathering, material painting, and asset preparation, fragmenting the pipeline and reducing efficiency.
Solution
Mixar is a complete 3D editor built on Blender, rebuilt around AI to automate the grunt work of 3D content creation. It introduces background agents, called Mixie, that can be dispatched to handle unwrapping, baking, retopologizing, or asset generation while the artist continues working in the same scene. The editor assigns each object to a separate sub-agent for modeling or texturing, allowing parallel work streams that all return results to the same Blender scene. Mixar also includes layer-based texturing that works like Photoshop for materials, with masks, blend modes, and procedural inputs, plus an in-canvas moodboard for generating references and prompting scene changes without leaving the editor.
Target Audience
Primary users are indie game studios, technical artists, concept and environment artists, character artists, and solo creators, as well as teams in event and exhibition design, architecture and interiors, warehouse and facility planning, and real estate concepting.
Features
- Background agents (Mixie) that autonomously complete UV unwrapping, baking, retopologizing, and asset generation while the user keeps working
- Parallel sub-agent architecture where each object is assigned to a separate agent for simultaneous modeling or texturing, with results merged into one Blender scene
- Layer-based texturing system with PBR effects, masks, blend modes, and procedural inputs that remain editable by hand after AI generation
- In-canvas moodboard for generating references, pinning style anchors, and prompting scene changes directly inside the editor
- Open-source codebase available via GitHub, allowing users to inspect, run, and build on the software
- Bring-your-own-key API integration with pay-per-use pricing and the ability to swap AI providers at any time