Kazama-Suichiku/Houdini-Agent

Houdini Agent - DCC Asset Manager with AI capabilities

What it solves

Houdini Agent is an AI assistant designed to automate and streamline the complex procedural 3D workflow in SideFX Houdini. It removes the manual effort of building node networks, writing VEX/Python code, and sourcing 3D assets by providing an autonomous agent that can directly manipulate the software's environment.

How it works

The project operates as a standalone desktop application that connects to a running Houdini session via a lightweight bridge. It uses an autonomous agent loop powered by OpenAI's Function Calling protocol, allowing the AI to plan tasks, execute tools, and iterate based on results.

Key mechanisms include:

  • Agent Modes: It offers "Agent mode" for full scene mutation, "Ask mode" for read-only analysis, and "Plan mode" for creating structured execution plans with DAG flow diagrams before execution.
  • Tool Integration: The agent has access to over 50 tools for node operations (creating, connecting, and modifying nodes), geometry analysis (via a "Skills" system of pre-optimized Python scripts), and system-level execution (Python and Shell).
  • 3D Generation: It integrates the Meshy AI API to generate textured 3D models from text or images, which are then automatically imported into Houdini as GLB assets with PBR shaders.
  • Memory & Context: A brain-inspired three-layer memory system (episodic, semantic, and procedural) allows the agent to learn from past interactions, while user-defined context rules provide persistent instructions.

Who it’s for

Technical artists, 3D generalists, and Houdini users who want to accelerate their procedural generation workflows, automate repetitive node setups, or integrate generative AI 3D assets directly into their scenes.

Highlights

  • Autonomous Loop: Capable of multi-turn tool calling and complex task decomposition into tracked subtasks.
  • Meshy AI Integration: Full pipeline for text/image-to-3D, retexturing, and remeshing with an interactive concept gallery for human-in-the-loop selection.
  • Deep Integration: Can read node networks, execute VEX wrangles, and navigate to specific nodes via clickable paths in the chat.
  • Advanced UI: A modern QML/Qt Quick "Mono Editorial" interface featuring real-time token analytics, streaming code previews, and a cloud asset library.
  • Extensible: Supports a plugin system for community extensions and a wide array of AI providers (DeepSeek, OpenAI, Claude, etc.) via OpenRouter or custom endpoints.

Related

  • Project
    wassermanproductions/motion-previs-studio

    A desktop app for filmmakers to extract pose, depth, and camera movement from reference videos to create control bundles for AI video generation pipelines.

  • Project
    ganbo-gab/open-storyboard-canvas

    Open Storyboard Canvas is an open‑source, cross‑platform desktop app (Tauri + React + Rust) that provides a visual node canvas for AI‑generated images, videos, and 3‑D storyboard scenes. It includes a chat‑based Canvas Agent (beta) that can create, edit, and track generation tasks, supports multiple providers (OpenAI‑compatible, Claude, Dreamina CLI), and offers director‑studio tools, prompt libraries, bulk import, and detailed logs. Licensed MIT, built on the Storyboard‑Copilot upstream project.

  • Project
    ixaxaar/pytorch-dnc

    A PyTorch library implementing Differentiable Neural Computers (DNC), Sparse DNC (SDNC) and Sparse Access Memory (SAM) with install instructions, API usage, debugging support, and example training tasks (copy, addition, argmax).

  • Project
    leamsigc/ShortsGenerator

    An automated pipeline for creating short-form videos that handles AI script generation, stock footage sourcing, TTS voiceovers, and social media scheduling.