Nodes/comfyui_LLM_party/URL 2 IMG🐶
ComfyUI Node

URL 2 IMG🐶

Pull an external image straight into your LLM Party graph

By heshengtao·Created 2 years ago·Updated 7 days ago· 2,321
URL 2 IMG🐶
    • file_path
    • img
    • log
    url
    file_name
    is_enabletrue

    Another one carrying the pack's little dog emoji in its display name - "URL 2 IMG 🐶" - and, like the other emoji'd nodes here, that's just house style, not a hint about what it does. What it does is straightforward: hand it a direct link to an image, and it downloads it, saves it locally, and gives you back both a usable IMAGE and the local file path.

    The obvious use in an agent context: your LLM found an image URL somewhere - a web-search tool result, a link a user pasted into chat - and now you need that image actually inside your ComfyUI graph to do something with it, not just a URL string sitting in text. URL2IMG is the node that closes that gap.

    Inputs and outputs that matter

    • url - a direct link to an image file.
    • file_name - what to save it as locally.
    • is_enable - the pack-wide skip switch.

    Three outputs: file_path, the saved location on disk; img, the image itself as a normal IMAGE, ready for the rest of your pipeline; and log, a status string - useful for confirming the fetch actually succeeded, or seeing why it didn't, without the whole run erroring out.

    Installing it

    Search comfyui_LLM_party in ComfyUI Manager and install, or clone by hand:

    cd ComfyUI/custom_nodes
    git clone https://github.com/heshengtao/comfyui_LLM_party
    

    pip install -r requirements.txt inside your ComfyUI Python environment, then restart. This node needs outbound network access but no model or API key of its own - it just fetches a URL - though it ships bundled inside the full pack, which is a much larger install overall. If you only want API-based LLM calling, the README's only_api branch skips the heavier local-model dependencies.

    Common issues

    url needs to be a direct link to an image file, not a webpage that happens to display one - point it at a page instead of the actual image URL and you'll get back whatever content-type that page actually served, not a usable picture. Some image hosts also block hotlinking or require a browser-like request to serve the file at all, so a link that opens fine when you click it in a browser can still fail here.

    The other real trap, and it's worth checking before you build a whole workflow around this node: if you're running ComfyUI in a locked-down or offline environment, outbound network access to arbitrary URLs may not even be permitted in the first place. That's a network/infrastructure constraint, not something wrong with the node - confirm your executor can actually reach the open internet before assuming a failed fetch is a bug. And if you only need this capability available for an LLM agent to invoke on its own rather than wiring a fixed URL by hand, this pack's url2img_tool wraps the same underlying capability as a callable tool instead.

    Category大模型派对(llm_party)/转换器(converter)

    Inputs (3)

    NameTypeDefaultDescription
    urlSTRING
    file_nameSTRING
    is_enableBOOLEANtrue

    Outputs (3)

    NameTypeDescription
    file_pathSTRING
    imgIMAGE
    logSTRING