tigerowo/infinite-canvas

What it solves

Infinite Canvas is an open-source creative workbench designed for image, video, and audio generation. It solves the problem of fragmented workflows by integrating canvas-based orchestration, AI generation tools, reference image editing, a dialogue assistant, and a prompt library all within a single interface, making it easier to explore visual schemes and iteratively refine results.

How it works

The project uses an infinite canvas as the primary interface where users can drag, scale, and connect nodes. It integrates with OpenAI-compatible APIs for image and text generation. It also features a "Director's Stage" (导演台) for 3D scene arrangement—including characters, models, and panoramic environments—which can then be used to generate specific camera shots that are sent back to the canvas as image nodes.

Who it’s for

Visual creators, AI artists, and designers who need a continuous, non-linear workflow for generating and iterating on multimodal AI content.

Highlights

• Integrated AI Workbench: Combines image/video/audio generation, reference image editing, and prompt management in one place.
• 3D Director's Stage: Allows users to actually arrange 3D scenes and camera positions to automate the creation of prompt parameters for visual generation.
• Canvas-based Workflow: Supports multi-canvas projects, node connections, and custom workflows with variable forms.
• Canvas Assistant: An AI assistant that can converse about selected nodes and insert generation results directly back onto the canvas.
• Prompt Library: Includes a curated collection of hundreds of image prompts gathered from various open-source projects.