ComfyUI interface basics: read the canvas, nodes and queue

What you are looking at

ComfyUI is a visual editor for a workflow graph. A workflow is a sequence of operations represented by nodes on a canvas. Links carry typed data from one node’s output to another node’s input; node properties hold settings such as a prompt or image size. A typical image graph loads a model, encodes text, samples an image and saves a result. Other workflows can create video, audio or 3D outputs, and the graph differs by task. Official workflow concepts

Find the main interface areas

The current interface documentation describes a main menu, a central workflow workspace, left-side panels for assets, nodes, models, workflows and templates, bottom toolbar controls, and a right-side area for running work and viewing the queue. The exact arrangement and names can change between frontend versions or menu modes; use the settings and help in your installed version if a label differs. Official interface overview

• Canvas / workspace: the graph is arranged here. Pan and zoom to inspect it; selection and navigation shortcuts are configurable.
• Nodes: each box performs a job, such as loading a model, processing an image, or saving output. A missing node usually means the needed class is unavailable or did not load; it is not itself proof that a model file is missing.
• Links: follow the wires from inputs through processing to output. Connectors carry particular kinds of data, so an image socket and a model socket are not interchangeable.
• Models and templates: the model panel can show detected local model files. Built-in templates are starting graphs and may ask you to obtain their required models. A model family or filename alone does not establish compatibility; check the workflow’s documented pairing. Official model guide
• Queue / execution: submitting a workflow schedules its graph for execution. Queue state and runtime messages help distinguish waiting, running, completed and failed work.

Before the first run

1. Choose a workflow intended for your task and ComfyUI version. Prefer a built-in template or a source whose author identifies its required model files and custom nodes.
2. Read the graph from source to output. Confirm model loaders select files available to this ComfyUI instance, required input images are present, and the final branch reaches an output node.
3. Resolve missing nodes or files before repeatedly submitting. Do not install a package just because a node label looks unfamiliar; check its technical class and source.
4. Submit once, then inspect both the queue and the node marked with an error if it fails. A workflow is not proven successful because it appears on the canvas.
5. Open the output node or the documented output location for that workflow. Some node types save to task-specific subfolders; do not assume every output is in the same folder.

The current official shortcuts page lists execution, workflow save/load, settings and canvas navigation shortcuts, but key bindings can be customized. Open the Settings shortcut list in your own interface before relying on a shortcut. Official keyboard and mouse shortcuts

Evidence and limits

This is a reference summary based on the official pages linked below, checked on 2026-09-23. It does not include a screenshot from a particular release, claim a GPU run, or guarantee that controls appear in the same positions in Desktop, Cloud, legacy-menu or future versions. Use the in-app help and the documentation matching your installed frontend when details differ.