Go back

How to Prompt Bonsai Studio: A Beginner’s Guide to Danbooru Tags in Stable Diffusion

Thursday, August 14, 2025

If you’re new to Bonsai Studio and want clean anime images, the fastest path to good results is learning Danbooru-style prompting. Our backend uses a custom Stable Diffusion model which is trained to “understand” Danbooru tags, so short, precise tag lists often beat long sentences.

What Danbooru-style prompting means

Danbooru is an anime image tagging site. Tags are short, lowercase words with underscores instead of spaces, like blue_hair or school_uniform.

You can use natural language, but tag-style prompts give much more consistent control because the model was trained on them.

Prompts are split into:Positive prompt: what you want to see.Negative prompt: what you want to avoid.

Core prompting rules

  • Use lowercase tags separated by commas. Replace spaces with underscores.
  • Order matters. Put the most important concepts first.
  • Be specific but not overloaded. 10–30 well-chosen tags usually work best.
  • Avoid contradictions. Don’t ask for short_hair and long_hair together.
  • Start simple, iterate, and add tags only when needed.

Essential tag categories

(Pick one or two tags from each category to build a clear prompt)

Subject

Count and type

  • 1girl, 1boy, solo, multiple girls, group

Age/shape (keep it general and safe)

  • teen, young adult, adult

Shot type

  • portrait, upper body, full body, close-up

Face and hair

Hair color

  • blue hair, blonde hair, black hair, brown hair, pink hair, silver hair

Style

  • long hair, short hair, twintails, ponytail, bob cut, bangs, ahoge

Eyes

  • blue eyes, green eyes, heterochromia, glasses

Expression

  • smile, closed mouth, open mouth, blush, determined, smug

Gaze

  • looking at viewer, looking to the side, looking up, looking down

Pose and composition

Pose:

  • standing, sitting, running, jumping, crossed arms, hands on hips

Framing:

  • centered, from below, from above, low angle, dutch angle, profile, back view

Cropping:

  • upper body, full body, bust, cowboy shot, knees up

Focus:

  • depth of field, bokeh, focus on face

Clothing and accessories

Uniforms:

  • school uniform, sailor collar, blazer, necktie, ribbon

Casual:

  • hoodie, jacket, turtleneck, sundress, jeans

Styles:

  • kimono, yukata, gothic lolita, streetwear

Details:

  • hair ornament, hat, scarf, gloves, thighhighs, boots, sneakers, backpack

Background, setting, and mood

Indoors:

  • classroom, cafe, bedroom, library, office

Outdoors:

  • cityscape, street, park, field, beach, shrine, cherry blossoms

Weather/time:

  • sunny, overcast, rain, snow, night, sunset, golden hour

Lighting:

  • soft lighting, rim lighting, backlighting, volumetric lighting, dramatic lighting

Background complexity:

  • simple background, detailed background

Style hints

  • anime, cel shading, vibrant colors, pastel colors, high contrast, painterly

Art-like topics:

  • watercolor style, sketch, monochrome (use sparingly; Animagine is already anime-focused)

Quality boosters (use lightly)

  • masterpiece, best quality, high detail, high score, detailed eyes, detailed hair, intricate details

Negative prompt basics

Generic quality negatives:

  • lowres, blurry, bad anatomy, bad hands, extra digits, missing fingers, extra arms, deformed, mutated

Cleanliness:

  • watermark, signature, text, logo, copyright, jpeg artifacts

Composition control:

  • duplicate, two heads, out of frame, cropped, bad perspective

Background control:

  • busy background (if you want simple), simple background (if you want complex; put in positive instead)

If you want strictly safe-for-work, add:

  • nsfw, nudity, cleavage, underboob, see-through

Emphasis and weighting

You can also control the weight of the tags by using special characters.

(Use weights sparingly. Try moving important tags earlier before increasing weights.)

Parentheses emphasize:

  • (blue hair) or (blue hair:1.3)

Square brackets de-emphasize:

  • [twintails] or (twintails:0.8)

You can also emphasize a group:

  • ((blue hair, twintails):1.2)

A reliable prompt structure

Quality/style (optional):

  • masterpiece, best quality

Subject and shot:

  • 1girl, solo, upper body, looking at viewer, smile

Identity features:

  • blue hair, short hair, blue eyes

Clothing:

  • sailor collar, school uniform

Composition/pose:

  • centered, depth of field

Background/mood:

  • classroom, soft lighting

Example prompts you can copy

  • 1girl, fox ears, multiple tails, standing, looking at viewer, fox shadow puppet, gentle smile, closed mouth, night, starry sky, house, stone road, depth of field, dutch angle, sensitive, masterpiece, high score, great score, absurdres
  • 1boy, solo, white hair, donut hair bun, medium hair, blue eyes, empty eyes, expressionless, black gakuran, squat, head rest, blue sky, clouds, from below, masterpiece, high score, great score, absurdres

How to find the right tags

  • Use Danbooru’s tag pages to discover exact names. Replace spaces with underscores.
  • If a tag doesn’t work, try a simpler or more common synonym (e.g., twintails vs twin tails).
  • Look at related tags on Danbooru to expand your vocabulary for hair, clothing, and poses.

Troubleshooting common issues

Extra people appear:

Add solo and 1girl, and remove group/multiple tags.

Background is too busy

Swap 'detailed background' for' simple background', and add 'busy background' to negative.

Hair or clothing color is ignored

Move that tag earlier and/or weight it: (blue hair:1.3).

The face isn’t looking at you

Add 'looking at viewer' and centered.

Results feel noisy

Remove overlapping tags and reduce quality buzzwords; fewer, stronger tags beat many weak tags.

Do’s and don’ts

  • Do start minimal and iterate after each render.
  • Do keep your most important ideas at the front of the prompt.
  • Do use negatives to enforce cleanliness (no text, no signatures).
  • Don’t pile on conflicting attributes.
  • Don’t spam too many quality words; one or two are enough.
  • Don’t fight the model with long prose; use clear tags instead.

Conclusion

Mastering Danbooru-style prompting is the key to unlocking the full creative power of Bonsai Studio. By using clear, specific tags and refining your prompts step by step, you’ll be able to generate high-quality, stunning images even if you’re a beginner.

Ready to put your new prompting skills into action? Download Bonsai Studio today and start creating your own! Click the link below to get started and see what you can imagine with Bonsai Studio!

Start creating now.

Download Bonsai Studio (v0.3.7)