ComfyUI Node
Anima Prompt Studio
A ComfyUI node in HaDeS/Anima Pipeline with 14 inputs and 5 outputs.
Anima Prompt Studio
- clip
- positive
- negative
- positive_string
- negative_string
- info
◄model_variant▾►
◄style_mode▾►
◄booru_sanitizertrue►
◄t5_weight_multiplier1.00►
◄character_tag►
◄artist_tag►
◄positive_prompt1girl, solo, silver hair, cyan eyes, detailed anime illustration, soft rim lighting►
◄negative_promptworst quality, low quality, bad anatomy, bad hands, blurry, jpeg artifacts, artist name, watermark, signature►
◄artist—►
◄header_mode▾►
◄header_trigger▾►
◄artwork_title►
◄character—►
CategoryHaDeS/Anima Pipeline
Inputs (14)
| Name | Type | Default | Description |
|---|---|---|---|
| model_variant | COMBO | 5 options: Anima Aesthetic (Strip score_* & Slop Guard), Anima Aesthetic v1.0b (Keep masterpiece, best quality), Anima 2.9B / 3.8B Expanded (Keep Quality Tags), Anima Base 2B (Keep Quality Tags), Anima Turbo / Lightning (CFG 1.0 Optimized) | |
| style_mode | COMBO | 4 options: Standard Anime / Illustration, Flat Vector / Graphic (Non-Anime Dataset Header), Cinematic / Detailed Shading, Raw Pass-Through | |
| booru_sanitizer | BOOLEAN | true | Normalize Booru tags to lowercase, replace underscores with spaces (except score_*, *_up, safe_for_work, kaomojis), escape literal parens, order into canonical buckets, and preserve natural language capitalization. |
| t5_weight_multiplier | FLOAT | 1.000–2.5 | Scales T5XXL query-token emphasis weights (tag:weight) exiting Anima's llm_adapter bridge. |
| character_tag | STRING | Optional character and series tag(s) (e.g. frieren, sousou no frieren). Automatically normalized and placed in canonical character position. | |
| artist_tag | STRING | Optional artist tag(s) with optional weight (e.g. @zocehuy, (@zocehuy:1.2)). Automatically prefixed with '@' and placed in canonical artist position. | |
| positive_prompt | STRING | 1girl, solo, silver hair, cyan eyes, detailed anime illustration, soft rim lighting | — |
| negative_prompt | STRING | worst quality, low quality, bad anatomy, bad hands, blurry, jpeg artifacts, artist name, watermark, signature | — |
| clipopt | CLIP | Optional Qwen3/Anima CLIP to encode CONDITIONING directly with T5 weight scaling. | |
| artistopt | STRING | Optional external artist tag(s) input from selector or text nodes. | |
| header_modeopt | COMBO | Flat Vector / Graphic header format. Canonical = CircleStone Labs 3-line format: trigger on Line 1, optional artwork title on Line 2, prompt on Line 3. Dual Legacy = legacy ye-pop deviantart header. | |
| header_triggeropt | COMBO | Dataset trigger for Line 1 of the canonical header. | |
| artwork_titleopt | STRING | Optional artwork title / alt-text for Line 2 of the canonical header. Leave empty to omit Line 2. | |
| characteropt | STRING | Optional external character tag(s) input from selector or text nodes. |
Outputs (5)
| Name | Type | Description |
|---|---|---|
| positive | CONDITIONING | — |
| negative | CONDITIONING | — |
| positive_string | STRING | — |
| negative_string | STRING | — |
| info | STRING | — |