Get Clip Model Names (Texturaizer)

A small, single-purpose node: it hands you the names of two CLIP models, and per the pack's own naming it's specifically thinking about Flux's dual text-encoder setup. If you've built a workflow where Blender is the source of truth for which checkpoints and encoders get used, this is the node that reports which CLIP files it decided on, without you having to keep a dropdown in ComfyUI in sync with the addon's preferences by hand.

How it works

It's the same read pattern as the rest of the pack's "Get" family - point directory_optional at Texturaizer's export folder, or leave it blank and it checks the addon's configured global directory, or feed it an already-loaded dictionary via data_optional (typically chained from a Get Json Data node so the JSON only gets read once per run). From that data it pulls out two specific fields and reports them as plain values.

The outputs that matter

There are only two real outputs here, and they're what makes this node worth a dedicated slot rather than folding it into Get Model Name:

• clip_model_1 - the tooltip calls it "First flux CLIP model name."
• clip_model_2 - "Second flux CLIP model name."

Flux-class models run two text encoders side by side - CLIP-L handles short, tag-like description and T5-XXL (the much larger language model half of the pair) is what gives Flux its natural-language prompt comprehension. Both process your prompt and their outputs get combined, which is why a Flux loading setup needs two separate CLIP file names rather than one, and why this node exists as its own thing instead of a single "clip name" output. Wire both into whichever dual-CLIP loader you're using upstream of your text encode node.

Plus the standard data_hash - debug fingerprint only, per the author's tooltip, not meant to be wired anywhere.

Installing it

ComfyUI Manager → search Texturaizer → Install → restart, or clone it by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/LatentSpaceDirective/ComfyUI-Texturaizer

Restart ComfyUI. No extra Python packages called out in the README, and the pack doesn't bundle any CLIP weights itself - it's reporting filenames, not downloading anything. You still need Flux's actual CLIP-L and T5-XXL files sitting in your clip folder for whatever it reports to resolve to something real.

Where people get burned

The gap between "this node returns a name" and "ComfyUI can load a file by that name" is the whole failure mode here. If the string that comes back doesn't exactly match a file you actually have installed, your downstream CLIP loader will fail to find it - this node doesn't validate that the model exists, it just reports what Blender's scene data says it should be. And as with every other getter in this pack: it's built to run downstream of the Blender addon. Drop it into a workflow with nothing feeding directory_optional or data_optional, and it's reading from a "global directory" that only means something once the Blender side has been configured and has actually exported. If you're building a Flux workflow from scratch without Blender in the loop, you're better off using a standard dual-CLIP loader node directly rather than routing through this one.