#1/2 How to Prompt the Anima AI Model: A Complete Guide
6
48
What is Anima?
Anima is a 2 billion parameter text-to-image model created via a collaboration between CircleStone Labs and Comfy Org. It is focused mainly on anime concepts, characters, and styles, but is also capable of generating a wide variety of other non-photorealistic content. The model is designed for making illustrations and artistic images, and will not work well at realism. It is trained on several million anime images and about 800k non-anime artistic images, with no synthetic data used for training.
Prompting Basics
Anima is trained on Danbooru-style tags, natural language captions, and combinations of tags and captions. Use lowercase for tags and spaces instead of underscores score tags are the only exception and do use underscores.
Recommended positive prefix: > masterpiece, best quality, score_7, safe,
Recommended negative prompt: > worst quality, low quality, score1, score2, score_3, artist name
Tag Order
Structure your tags in this order: [quality/meta/year/safety tags] [1girl/1boy/1other] [character] [series] [artist] [general tags] Within each section, the tags can be in arbitrary order.
Quality Tags
There are two quality tag systems. Human score-based tags: masterpiece, best quality, good quality, normal quality, low quality, worst quality. Aesthetic model-based tags (PonyV7 style): score9, score8, ..., score_1. You can use either system, both together, or neither all combinations work.
Time Period Tags
You can specify a year with tags like year 2025, year 2024, or use period descriptors: newest, recent, mid, early, old.
Artist Tags
Prefix an artist's name with @ for example, @ big chungus. The effect will be very weak if you don't include the @ symbol.
You can also stack multiple artists and weight their influence using the :weight syntax:
(@zocehuy:1.2), (@zomzomzomsauce:1.0)
A weight above 1.0 strengthens that artist's influence. This lets you blend styles the dominant artist leads the aesthetic while the second one contributes texture, color mood, or line quality. Experiment with values between 0.8 and 1.4 for best results.
Safety Tags
Use safety tags to control content: safe, sensitive, nsfw, explicit. Use appropriate safety tags in both the positive and negative prompts, alongside writing sufficiently detailed prompts, to avoid undesired content.
Natural Language Prompting Tips
Follow standard English capitalization rules for character and series names. If using pure natural language, more descriptive is better aim for at least 2 sentences, as extremely short prompts can give unexpected results. You can mix tags and natural language in arbitrary order.
When prompting for characters, name the character first, then describe their basic appearance. This is especially important when prompting for multiple characters if you just list off character names with no description of appearance, the model can get confused.
One powerful technique is layering your prompt in logical blocks, moving from global quality → subject → specific details → atmosphere → mood tags. This mirrors how the model was trained and gives it clear anchors at each level of abstraction.
