Anima Artist Tag Selector

If you run the Anima model, you already know the drill: artist tags are the single most powerful thing in the prompt. "Artist tags is pretty overpowered on this model" is one of the community's most repeated lines about it, and it's true. The problem is finding the artists worth using - there are over 40,000 of them baked into Anima's training, and nobody can name that many from memory. That's the whole job of the Anima Artist Tag Selector.

It's part of the Anima-Tools pack (nregret/Comfyui-Anima-Tools), which lives under the AnimaArt category in the node menu. The selector ships the artist database as a bundled data file derived from ThetaCursed's Anima Style Explorer - 40,000+ Danbooru artists with preview images, post counts, and a uniqueness score. In the node you get a visual grid of 3:4 preview cards you can search and paginate, sorted by works count, uniqueness score, or A–Z. Click artists, they land in the artist_tags field, wire the output into your positive prompt, done. No API keys, no calls anywhere - the "premium" is just a big local database with a nice front end.

What the node actually does to your tags

The interesting part is what happens on the Python side, because it's opinionated in a way that saves you from yourself. Anima expects artist styles in @name format (think @dairi, @wlop). Type dairi, by dairi, or even paste @dairi - the node normalizes all of them to @dairi, stripping stray by prefixes and duplicate @ signs. There's also a _raw_: escape hatch: prefix a tag with _raw_: and it's passed through untouched, normalization skipped, for when you genuinely want the raw text.

The inputs you'll actually touch:

• artist_tags - the comma-separated list, filled automatically by the panel. You can also type into it directly.
• mode - append (default) or override. With append, your artists get prepended to whatever comes in through opt_prompt; with override, the selected artists replace it entirely.
• opt_prompt - optional; wire an existing prompt string in if you're merging with another prompt source.

The single output is text (a STRING), which you feed into a CLIPTextEncode node - or into one of the pack's own assembler nodes.

One quirk worth knowing: the output ends with a trailing ", " (the author's deliberate choice so the artist block leads and whatever follows gets cleanly separated). That's fine when the next thing is a normal text encoder, but it means chaining raw selector outputs together produces double commas. That's exactly the problem the Selector+ variants and Anima Prompt nodes exist to fix.

Installing and what to expect

This whole pack has zero Python dependencies - no pip install, no requirements.txt, just frontend JavaScript and stdlib Python. Either install through ComfyUI Manager (search "Anima Tools"), or:

cd ComfyUI/custom_nodes
git clone https://github.com/nregret/Comfyui-Anima-Tools

Then restart ComfyUI. No model files to download separately; the 40k-artist database ships inside the pack. One caveat: the preview thumbnails load from a CDN (jsDelivr/GitHub Raw/Statically), so the visual grid needs internet even though the node itself works offline. And if a node doesn't show up after install, it's almost always just that you haven't restarted ComfyUI yet.

The README hasn't fully caught up with the code - it documents this node's artist/character/clothing siblings but not the newer background and pose selectors. The code is the source of truth, and it's solid.

If you're new to Anima, budget an hour clicking through this grid before you prompt anything serious. Find a dozen styles that click, save the tags, and build from there - it's the difference between "Anima looks bad" and "Anima is the best anime model I've run."