Nodes/ComfyUI Ino Nodes/Ino Get Lora Download Config
ComfyUI Node

Ino Get Lora Download Config

Pick a LoRA from a dropdown and get a download config string — no manual file hunting

By nobandegani·Created about a year ago·Updated 2 months ago· 6
Ino Get Lora Download Config
    • ModelConfig
    enabledtrue
    model

    Finding the right LoRA file usually means a trip to Civitai, a check that the download URL is still alive, and a guess about whether it's actually the file the workflow expects. Ino Get Lora Download Config kills that whole chore: you pick a LoRA from a dropdown, and it returns a JSON config string that tells the pack's downloader exactly what to grab and from where.

    It's part of the InoNodes "download from inside the graph" philosophy. This node is the pick the LoRA step; it doesn't touch the network itself. Its one output, ModelConfig (a STRING), is meant to feed InoHandleDownloadModel, which reads the config, picks the host (HuggingFace, S3, or Civitai), downloads the file into the right model folder, and hands you back a path you can wire straight into a LoRA loader.

    How it works

    The dropdown is populated from a CSV that ships inside the pack - data/files/lora_files.csv. Pick a row's name and the node serializes that whole row into JSON. Peek at the CSV columns and you'll see the config shape: name, weight_type, host, model_type, model_subfolder, repo_id, filename, subfolder, repo_type, revision. That's the same shape every "Get ... Download Config" node produces, which is exactly why one downloader node can handle all of them.

    Inputs and outputs

    • enabled - the pack-wide toggle. Off just returns an empty string, so you can gate this node without deleting it.
    • model - the dropdown of LoRAs from the bundled CSV.

    Output: ModelConfig - the JSON string that feeds the download chain.

    Installing it

    Part of ComfyUI-InoNodes (publisher Inoland). Easiest via ComfyUI Manager: search "ComfyUI Ino Nodes", install, restart. Manual:

    cd ComfyUI/custom_nodes
    git clone https://github.com/nobandegani/ComfyUI-InoNodes
    cd comfyui_ino_nodes
    pip install -r requirements.txt
    

    Then restart ComfyUI. The pack is built entirely on the V3 node schema, so it wants a recent ComfyUI (v0.18.1+); on an old build the nodes won't even register. inopyutils comes with the install. No API keys needed for this node itself - but if the LoRA you picked hosts on Civitai and it's gated, the download step may want a CIVITAI_TOKEN env var.

    Common issues

    The most common mistake is expecting this node to download. It produces a config string - if you wire it to nothing, you get a JSON blob and no file. That's by design; the download is the next node's job.

    Second, the dropdown is only as big as the shipped CSV. The bundled list skews toward LoRAs the author's own pipelines use - the WanVideo lightx2v LoRAs from Kijai's repo are in there, for example. If your LoRA isn't listed, there's no refresh button to find; the list is static until the pack updates it. For anything off-list you're better off with a config node that takes free-form fields, or just a plain LoRA loader. And when the config flows through to the loader, remember the strength split: model and CLIP strengths are separate knobs, and 0.7–0.9 is a saner starting point than maxing both.

    CategoryInoModelHelper

    Inputs (2)

    NameTypeDefaultDescription
    enabledBOOLEANtrue
    modelCOMBO14 options: kji-lightx2v-I2V-14B-480p-cfg-step-distill-rank128-bf16, kji-lightx2v-I2V-14B-480p-cfg-step-distill-rank16-bf16, kji-lightx2v-I2V-14B-480p-cfg-step-distill-rank256-bf16, kji-lightx2v-I2V-14B-480p-cfg-step-distill-rank32-bf16, kji-lightx2v-I2V-14B-480p-cfg-step-distill-rank4-bf16, kji-lightx2v-I2V-14B-480p-cfg-step-distill-rank64-bf16, +8

    Outputs (1)

    NameTypeDescription
    ModelConfigSTRING