Nodes/NodeGPT/DisplayTextAsImage
ComfyUI Node

DisplayTextAsImage

When the graph needs your LLM's words as pixels

By antonym-git·Created 3 years ago·Updated 2 years ago· 355
DisplayTextAsImage
  • text

    Every now and then a node exists because of a quirk of the host app, and this is one of them. ComfyUI is an image engine; its output nodes are wired to display images, and a plain text dump doesn't always survive the journey. DisplayTextAsImage solves that the blunt way: it takes the TEXT your NodeGPT agents produced and renders it as an image you can actually see in the UI. Same idea as the pack's DisplayString node, different medium - this one draws the text onto a canvas so it flows through ComfyUI's normal image-preview path.

    It's an output node, so there's nothing to wire onward: one input, text (TEXT), and that's the whole node. You drop it at the end of a chain, right after a Chat node or the llava captioner, and whatever came back shows up as a rendered image in the node. Handy when the transcript is long, the terminal's scrolling out of reach, and you want the result pinned on the graph itself.

    When you'd actually reach for it

    Honestly? This is a nicety node, not a workflow pillar. It shines as a debugging aid in this specific pack, which was famously rough around the edges: agents that echoed their input, chats that refused to terminate, output that lived in the terminal by default. Turning the reply into an image means it lands in the same place as all your other previews, so you can glance at the whole graph and see what the model produced without alt-tabbing. It's also a neat demonstration that in ComfyUI, anything can become pixels if you try hard enough.

    Installing it

    The pack's install is the standard custom_nodes dance:

    cd ComfyUI/custom_nodes
    git clone https://github.com/xXAdonesXx/NodeGPT
    

    Restart ComfyUI and let it auto-install requirements on startup. The catch, same as every node in this pack: as of this writing the repo returns 404 and the author's GitHub account is deleted, so that clone fails and ComfyUI Manager can't install the pack either. The surviving fork (Guizmus/NodeGPT) only carries the pack's earliest text-gen build - not the AutoGen-era nodes. So if you find this node in a saved workflow from 2023, it's a fossil you're reading about, not something you're installing fresh today. The functionality it provides - text as a viewable image - lives on in any number of maintained packs if you actually need it.

    CategoryAutoGen

    Inputs (1)

    NameTypeDefaultDescription
    textTEXT

    Outputs (0)

    No outputs