Skip to main content

Uncensored — This model declares no content filter. SpicyAPI forwards the request without adding a platform content scan; legal and consent duties remain with the customer.Browse every model at this filtering level
circlestone-labs/anima-turbo/edit

Turn a photo or an existing illustration into an anime illustration on Anima Turbo. A 0 to 1 strength control sets how much of the source survives: around 0.75 keeps the composition and redraws the style, and nothing is filtered.

From $0.012/image
Run model

Workbench

Input

Image editing
292/5000
Describe the whole picture you want back, not just the change: who is in it, what they wear, where they are. Tags, sentences or both. at least 1 characters · at most 5000 characters
1/1

Drop files or media links, paste, or browse · 1 file · Max 10 MB · JPG · PNG · WEBP · GIF

  • 1
1 / 1 files
The picture to redraw. The result keeps its aspect ratio; the resolution tier sets how many pixels come back.
affects price
1k returns about one megapixel and is the default; 1.5k returns about 2.4 megapixels and costs more. The aspect ratio always follows the source.
Transformation strength
How far to move from the source, 0 to 1; 0.8 by default. At 0 the source returns untouched, at 0.3 a photo is still a photo, around 0.75 it becomes anime with the same composition, and at 1 the source is ignored. no lower than 0 · no higher than 1
Seed
Keeps results comparable while you adjust strength. It does not guarantee the exact same pixels on every run. no lower than 0 · no higher than 2147483647
Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run
singer-photo-to-animeExample result
Animate
Edit
Split into layers

Related models and next tasks

Missing the model you need?

Tell us what to add and where you would use it. Requests go to the people who decide what gets onboarded next.

Anime art in seconds, from tags or from plain sentences

Anima Turbo by CircleStone Labs is a 2-billion-parameter anime and illustration model, built on NVIDIA Cosmos-Predict2 with a Qwen3 text encoder, in its distilled Turbo build for 8 to 12 sampling steps. In our 60 test runs an image took 2 to 16 seconds of model time, 52 of them under ten. It reads Danbooru-style tags, plain sentences and mixes of both, draws at about one megapixel (1k) or about 2.4 megapixels (1.5k) in 15 aspect ratios, and restyles photos. The publisher says Turbo is only slightly worse than its Aesthetic build on average while being much faster. Nothing on this endpoint filters the prompt or the result: our strongest content probe rendered as written on the first try.

Open the playground
Anime art in seconds, from tags or from plain sentences
Model
Anima Turbo by CircleStone Labs, 2 billion parameters, built on NVIDIA Cosmos-Predict2 with a Qwen3 text encoder
Speed
2 to 16 seconds of model time per image in our runs; 52 of 60 finished in under 10
Sizes
15 aspect ratios. 1k: 1024 x 1024 at 1:1, 1360 x 768 at 16:9. 1.5k, generated natively: 1536 x 1536 and 2048 x 1152
Prompt
Up to 5,000 characters of tags, sentences or both; no negative prompt
Editing
One source image, strength 0 to 1 (default 0.8); the result keeps the source's shape
Content filter
None; our strongest content probe rendered as written on the first try
Two ways to prompt
Tags for control, sentences for atmosphereWe ran one rainy-cafe scene twice on the same seed, once as a tag list and once as four sentences. Both gave a clear picture. The tag version drew the window with a city behind it; the sentence version leaned into the warm interior light and the rain streaks it described. Pick tags when you want to pin details, sentences when you want a mood.
Billing
The tier sets the price, nothing else doesEach image is billed by its resolution tier: the 1.5k tier costs more than 1k, while the aspect ratio does not change the price and an image edit costs the same as text-to-image. Failed requests are not charged. Current prices are on the model page, so draft at 1k and spend the larger tier on the keepers.
Limits
What it will not doAnima is an anime and illustration model, not a photorealistic one, and it renders only a word or two of text. There is no negative prompt, and one request returns one JPEG image. Editing restyles a picture; it does not follow instructions such as a new outfit or day to night. For a specific house style from a LoRA file, use Anima Turbo LoRA.

Prompting Anima Turbo

Change the whole look with a few tags

Change the whole look with a few tags

A handful of tags is enough to change the whole look of an Anima Turbo image. We fixed one character and one seed, a silver-haired woman in a trench coat on an autumn street, and changed only the style tags. “retro artstyle, 1990s (style), anime screenshot” gave a clear 90s cel look, “(chibi:2)” gave a chibi, and “monochrome, greyscale, manga, screentone” gave black-and-white manga. Watercolor and pencil-sketch tags changed the rendering only slightly. Weights such as (chibi:2) work, and they need higher values than you may be used to from SDXL.

Tags and sentences both work

Tags and sentences both work

Anima Turbo accepts tags, sentences and mixes of the two. In our cafe test the tag prompt and the four-sentence prompt each produced a usable picture of the same scene. The publisher recommends at least two sentences when you write in natural language, since very short prompts drift. A mix is fine too: open with the quality tags, then describe the scene in a sentence or two.

  • Order the tags the way the model expects

    Tag order matters to Anima. The publisher's convention is lowercase, spaces instead of underscores, and this order: quality and rating tags, head count (1girl, 1boy, no humans), character, series, artist, then general tags. A good opening is “masterpiece, best quality”, optionally followed by “score_7”.

  • Let the rating tag set the tone

    The rating tag is the main control over how far a picture goes, because there is no negative prompt. The model knows safe, sensitive and nsfw, plus one stronger tag. Every example on this page starts with safe, and nothing on the endpoint filters the result either way.

  • Change the seed when a corner scribble appears

    A new seed fixes a stray corner scribble. Anima sometimes draws a signature-like scribble or a line of pseudo-text in a corner, a habit learned from artwork; we saw it in roughly one image in five, and no negative prompt can suppress it. Seeds make runs comparable, but the same seed will not always give identical pixels. For an artist's style, put @ in front of the name, or the effect is weak.

Who uses Anima Turbo

  • A photo, redrawn as clean anime

    Creators with photos to restyle

    A photo, redrawn as clean anime

    Send a photo to the edit endpoint with a prompt that describes the same scene. Our stage photo of a singer at a microphone came back at strength 0.75 as clean anime with the same composition: dress, microphone and pose all stayed in place. Describe what is in the picture, not what you want to change.

  • Find the strength before you rewrite the prompt

    Anyone tuning an edit

    Find the strength before you rewrite the prompt

    Strength decides how much of a photo survives an edit. On one photo with one prompt and seed, 0.3 was still a photo, 0.6 already read as an illustration with the same details, 0.75 was clean anime with the same composition, and from 0.9 the details started to change. At 1.0 the source was ignored, and at 0 it comes back untouched. The set shows the source, then 0.3, 0.6, 0.75 and 0.9.

  • One model, a wide range of scenes

    Illustrators and story teams

    One model, a wide range of scenes

    One endpoint covers many kinds of scene. It gave us a couple walking a beach at sunset, an armoured knight on a cliff, a man in a suit on a night train, a cyberpunk alley with no characters in it, a man pouring tea on an autumn porch and a lantern-lit staircase in a tall 1:3 frame. Fifteen aspect ratios mean the frame can match the page, banner or phone screen you are drawing for.

  • Finished-size frames on the 1.5k tier

    Art and publishing teams

    Finished-size frames on the 1.5k tier

    The 1.5k tier is for frames you will publish. The bartender picture is 16:9 at 2048 x 1152, written as three sentences and generated natively, not upscaled. It costs more than a 1k image, so draft at 1k and rerun the one you like at 1.5k; expect a close cousin of the draft, not a pixel copy.

From first prompt to finished frame with Anima Turbo

  1. Open with quality and rating

    Start the prompt with “masterpiece, best quality” and a rating tag such as “safe”. It is the cheapest way to steer both quality and tone.

  2. Describe the scene

    Write tags in the publisher's order, or at least two plain sentences. Mention adults as adults, and say what the light and the setting are doing.

  3. Draft at 1k in the final ratio

    Composition shifts between 1:1 and 21:9 and the ratio does not change the price, so draft in the shape you will publish.

  4. Move the style, then the seed

    Add a style tag or two, and use a new seed whenever a corner scribble shows up. Change one thing per run so the comparison is honest.

  5. Restyle photos, then finish at 1.5k

    For existing pictures use the edit endpoint and start at strength 0.75. When a draft is right, rerun it on the 1.5k tier.

Anima Turbo prompts worth copying

  • A style swap in one line

    Keep the character tags and change only the style tags. This one gave a clear 90s cel look on our seed.

    masterpiece, best quality, score_7, safe, retro artstyle, 1990s (style), anime screenshot, 1girl, solo, adult woman, mature female, short silver hair, blue eyes, beige trench coat, scarf, standing on a city street in autumn, falling leaves, looking at viewer, soft smile

  • A scene in sentences

    Four sentences, written the way you would brief an illustrator. The rain on the glass and the warm light both came through.

    masterpiece, best quality, safe. An anime illustration of a man in his thirties sitting by a cafe window on a rainy afternoon. He has short black hair and glasses and wears a grey turtleneck, holding a cup of coffee in one hand while reading a paperback. Warm light from inside the cafe falls across the table, and raindrops streak the glass behind him.

  • A photo repainted as anime

    For the edit endpoint at strength 0.75. Describe the whole picture, because the model restyles the frame rather than following edit instructions.

    masterpiece, best quality, safe, anime illustration, 1girl, solo, adult woman, mature female, short dark wavy hair, dark green velvet dress with long sleeves, holding a vintage microphone on a stand, jazz club stage, warm spotlight from above, dark background, soft smile, looking to the side

More Anima Turbo questions

Use it as a quick sketchbook

Open with “masterpiece, best quality” and a rating tag, describe the scene in tags or sentences, and draft at 1k in the ratio you will publish.

Change one thing per run: a style tag, the seed or the edit strength. When a draft is right, rerun it on the 1.5k tier. If a signature-like scribble shows up in a corner, change the seed.

All Anima models2