Anima 2B Prompt Template 2

Anima's model card hands you a specific prompt shape: quality and meta tags up front, then character, series, artist, and general tags, with an era and score markers sprinkled in. Anima2BPromptTemplate2 is that recommended skeleton, turned into a node with a form field for every part and a toggle to switch each one off. If the pack's Template 1 is "which artists," this is "the rest of the prompt, in the right order, every time."

How it works

This one is a string assembler with opinions. It builds the fixed section of an Anima prompt in a strict order:

character_from → style_tags → year → quality_tags → score_a/score_b → custom_text

...then prepends template1_text if you wired one in, and appends upstream_text if you didn't leave it empty. The formatting is fussy on purpose - it normalizes comma spacing and makes sure each section ends with the right punctuation so the final string doesn't read like a robot glued sentences together. Every section has a matching *_switch boolean. That's the whole feature: leave a section's switch on to include it, flip it off to drop it from the assembled prompt entirely.

The inputs that matter

There are a lot of fields, but only a handful you'll actually touch:

• style_tags - the actual style words, and the one you'll edit constantly. The default ("crisp lines, smooth semi-realistic shading, ray tracing, translucent hair...") is a decent painterly starting point, but you'll be replacing it with tags you picked from the Style Explorer or animadex.
• character_from - defaults to hololive. This becomes The character is from: hololive. at the front of the prompt, which is exactly the kind of series-anchor phrasing Anima understands. Leave it blank and it's skipped automatically, even with the switch on.
• year - era matching, e.g. year 2021, . Some artists have distinct styles across eras - the README's example is @happoubi jin - so this is how you pin the timeline instead of writing "2021" into a free-text blob.
• quality_tags - newest, masterpiece, best quality by default. Fine to keep.
• score_a / score_b - the Pony-style score tags. This is the one field I'd actually change.

The score tag caveat

The default is score_a=8, score_b=7, which emits score_8, score_7.. Anima was trained on these tags and the card recommends score_7, but the community consensus is that the high end does something specific: score tags narrow the output distribution toward a generic polished look, and 8/9 in particular steer toward the Pony/e621-adjacent aesthetic. The safe read: if your outputs start smelling like generic Pony, drop score_a to 7 or flip score_a_switch off. And if you're already running a strong artist tag, the community advice is to drop the quality/score noise entirely - the artist tag is doing more work than masterpiece ever will.

Wiring it up

The intended chain is Template 1 → Template 2 → CLIP Text Encode. Feed Template 1's text output into template1_text here, and Template 2 assembles [artists] [character/series] [style] [year] [quality] [scores] in one pass. Alternatively run it standalone with upstream_text carrying your character prompt and take text straight to a positive prompt. Either way: string in, string out, nothing heavier.

Install

Same pack as Template 1 - this is why you installed it. No Python dependencies (the requirements file explicitly declares none), so:

cd ComfyUI/custom_nodes
git clone https://github.com/mizukir0418-gif/ComfyUI-Anima-Style-Explorer-Prompt.git

Restart ComfyUI and both templates appear under PromptHelper. ComfyUI Manager works too - search the pack title "Anima Style Explorer." The optional ~1.6 GB sidebar database is for the style browser, not for any of the prompt nodes.