TextNode

A whole node dedicated to passing a string through, unmodified. Which sounds like a joke until you see the sibling node it feeds: IMAGDressingNode takes its prompt as a typed TEXT input, and in the workflows this pack ships, that prompt comes from TextNode. Every one of the pack's five example graphs - base, controlnet, ipa_controlnet, cartoon, inpainting - routes the prompt through this little box first.

The entire mechanism is one function: encode(text) returns (text,). That's it. It's the ComfyUI equivalent of a straight wire, and there's no model, no GPU work, and nothing to download.

Input and output

• text (STRING, multiline) - your prompt, in a resizeable box.
• Output: TEXT - the same string, now carrying ComfyUI's typed TEXT type.

That type matters more than it looks. In current ComfyUI, TEXT is a distinct type from STRING - the type system that lets text be generated, batched, and wired through the graph like an image. A plain STRING output won't plug into IMAGDressingNode's prompt input, which is exactly why this node exists: it's the pack author's bridge between the new typed-TEXT convention and a simple editable box. The widget also declares dynamicPrompts, so ComfyUI's {red|blue} random-variation syntax is honored when the value flows through it - handy for generating a few outfit-description variants off one node.

Honest truth: you can also just type straight into the prompt box on IMAGDressingNode and skip TextNode entirely. It's not required. Where it earns its keep is when you want to branch, swap, or automate the prompt without touching the big node's widget - a second prompt source you can bypass or route from elsewhere in the graph. For a beginner following the pack's own workflows, it's just "the box you type your description into."

Installing

There's nothing separate to install - TextNode ships inside the same AIFSH/IMAGDressing-ComfyUI pack as IMAGDressingNode. Install the pack (ComfyUI Manager, search "IMAGDressing", or clone https://github.com/AIFSH/IMAGDressing-ComfyUI into custom_nodes/, then pip install -r requirements.txt and restart) and TextNode appears under the AIFSH_IMAGDressing category.

The only gotcha is type discipline: because it outputs TEXT, it only connects to inputs that accept TEXT. If you try to wire it into a node whose prompt field is a plain string widget, the connection won't take - that's the type system working as designed, not a bug in the node. Keep it feeding IMAGDressingNode (or any other TEXT consumer) and it'll never surprise you.