How invoke-ai/invokeai works
InvokeAI is a locally hosted creative engine for generating and refining visual media. Users work in the web UI, create node workflows or edit on the unified canvas, and receive generated media that can be organized and revisited in boards and the gallery. The README documents these capabilities; the source tree identifies the API, invocation, model-management, and persistence subsystems. Several runtime connections are therefore shown only where documented, or left unlinked where sampled evidence does not establish their caller.
Main entry points: invocations Invocation nodes · src Web UI · boards.py Boards and gallery · run_app.py Web server
Components 17
Creative interface
- Web UI
invokeai/frontend/webv2/src - Unified canvas
invokeai/frontend/webv2/src/workbench/canvas-engine/engine.ts
Application API
- Web server
invokeai/app/run_app.py - API routes
invokeai/app/api/routers
Workflow generation
- Workflow management
invokeai/app/api/routers/workflows.py - Invocation nodes
invokeai/app/invocations - Diffusion pipeline
invokeai/backend/stable_diffusion/diffusers_pipeline.py - Text LLM pipeline
invokeai/backend/text_llm_pipeline.py
Models and integrations
- Model API
invokeai/app/api/routers/model_manager.py - Model manager
invokeai/app/services/model_manager/model_manager_default.py - Model configuration
invokeai/backend/model_manager/configs/factory.py - External generation
invokeai/app/services/external_generation/providers
Media workspace
- Boards and gallery
invokeai/app/api/routers/boards.py - Image records
invokeai/app/api/routers/images.py - Video features
invokeai/app/api/routers/videos.py - Application database
invokeai/app/services/shared/database/engines.py
Other components
- Artist
Connections 16
- Artist → Web UI: uses
- Web UI → Web server: uses
- Artist → Unified canvas: edits media
- Web UI → Workflow management: manages workflows
- Web server → API routes: serves
- Workflow management → Invocation nodes: runs nodes
- Invocation nodes → Diffusion pipeline: generates media
- Invocation nodes → Text LLM pipeline: runs text models
- Model API → Model manager: manages models
invokeai/app/api/routers/model_manager.py - Model manager → Model configuration: uses configurations
invokeai/app/services/model_manager/model_manager_default.py - Invocation nodes → External generation: supports providers
- Artist → Boards and gallery: organizes media
- Boards and gallery → Image records: manages images
- Artist → Video features: uses video models
- Boards and gallery → Application database: persists records
- Image records → Application database: persists records