Nodes/Anymatix/Anymatix Quadruple CLIP Loader
ComfyUI Node

Anymatix Quadruple CLIP Loader

Four text encoders in one CLIP object — ComfyUI stops at two, this doesn't

By Anymatix·Created about a year ago·Updated 2 days ago· 0
Anymatix Quadruple CLIP Loader
    • CLIP
    clip_name1
    clip_name2
    clip_name3
    clip_name4

    Most models you'll touch need one text encoder, or two if you're in Flux/Hunyuan territory. ComfyUI's stock loaders top out there. So when a workflow asks for four separate text-encoder checkpoints to be merged into a single CLIP object, you've left the mainstream - and you need a loader that ComfyUI core doesn't bother to ship. AnymatixQuadrupleCLIPLoader is that loader, and it does exactly what it says on the box: load four CLIP/text-encoder files and stack them into one CLIP object that a single text-encode node can use.

    What it's really for

    Two things are true at once here. Functionally, the node is simple: four string inputs, each resolved to a path, and one call to comfy.sd.load_clip() with all four paths - plus the embeddings directory so your LoRA-trained embeddings still resolve. That's the entire mechanism; load_clip knows how to merge multiple encoder checkpoints into one object, and ComfyUI's own dual loader is just the two-file special case of the same call.

    The context is the pack. This is one of a family of loader twins (Triple, Quadruple, single, dual) that all exist for the same reason: Anymatix-exported workflows address their text encoders by URL, and the file may not exist when the graph is written. A dropdown can't point at a not-yet-downloaded file; a string can. A fetcher node downloads the encoder, hands the path over, and this loader resolves it. If you're not using Anymatix and you've somehow got a workflow with four text encoders, you can use it standalone too - just type or wire the four paths in.

    Inputs and output

    • clip_name1, clip_name2, clip_name3, clip_name4 - all STRING paths (or basenames) to the four encoder checkpoints. They resolve against your text_encoders folder.

    Output: CLIP, wired into a CLIP Text Encode node like any other. Order of the four names matters - they're merged in the order given, so keep the workflow's existing wiring and just swap in the loader.

    Install

    Shipped in the anymatix-comfy-nodes pack under the "Anymatix" category. ComfyUI Manager: search "anymatix-comfy-nodes", install, restart. Or by hand:

    cd ComfyUI/custom_nodes
    git clone https://github.com/Anymatix/anymatix-comfy-nodes
    

    then restart. You don't need the Anymatix app - the pack is published specifically so exported workflows run on stock ComfyUI.

    The honest caveat

    Unless you're reproducing an Anymatix workflow (or genuinely running a model with four encoders), you probably don't want this node - three is already exotic, and Quadruple is the "grab it when an exported graph asks for it" member of the family. If you're staring at one of these in a workflow you didn't build, the useful question isn't what it does, it's which four files the fetchers ahead of it are pulling down - that's where the real configuration lives.

    CategoryAnymatix

    Inputs (4)

    NameTypeDefaultDescription
    clip_name1STRING
    clip_name2STRING
    clip_name3STRING
    clip_name4STRING

    Outputs (1)

    NameTypeDescription
    CLIPCLIP