Same character. Any scene.
Under 5 seconds.

Generate a character once, then drop it into any prompt. The face stays consistent across every gen, with sub-second identity overhead per call.

ALL SYSTEMS OPERATIONAL ● ~5s / image ● from 2.4¢ / image
CONSISTENCY DEMO

One character · four scenes · same face

model: realistic-sharp-v1 · hd_portrait

sitting at a wooden cafe table, oat milk latte in hands, soft morning light through window, cream cardigan, candid photographic portrait

HD1 CREDIT

walking on a cobblestone street in autumn, camel wool coat, leather boots, scarf gently lifted by wind, golden hour, editorial

HD1 CREDIT
18+ NSFW

leaning against a rooftop railing at blue hour, black silk slip dress, city lights as soft bokeh background, glossy lips

HD1 CREDIT

standing on a beach at sunset, white linen dress flowing, hair tousled by sea breeze, warm golden hour light, cinematic

HD1 CREDIT

All four are the same character_id with different prompts, at hd_portrait resolution. Identity stays locked because we cache the face fingerprint per character.

MODEL VARIETY

Four image models · one API

see all models →
REALISTIC-SHARP-V1

Photorealistic portraits

The fastest of the photoreal models. 560 × 720 default; hd_portrait comes out at 1344 × 1728 after the always-on sharpening pass, in about 9 s. Identity stays consistent across every generation. 1 credit.

realistic-v2 close-up portrait, sharp fine detail
REALISTIC-V2

Sharpest and most realistic

Skin, hair and light that read as a photograph, with a sharp face by default. 896 × 1152, about 16 to 20 s per image. framing: "close_up" gives a true head-and-shoulders portrait. 2 credits.

semireal-v1 portrait in warm lamplight
SEMIREAL-V1

Semi-realistic illustration

A painted, semi-realistic look between photo and anime, with the same consistent face. 896 × 1152, about 10 to 13 s per image. 1 credit.

ANIME-V1

Stylized illustration

560 × 720 default, single-pass render, tag-style prompts. ~2-3s per image. Same six resolution presets as realistic-sharp-v1.

Plus video-v1 for image-to-video: same character, 5s motion clip, async + webhook.

Try the API in your stack

Create an account, mint an API key, ship your first gen in <10 lines of code.

Sign up free → Read the docs
TRY YOURSELF · LIVE

Generate against one of your characters

Images only here. Video generation and reference-image uploads are API-only — see the docs.
1 credit per gen · — credits left
CHARACTER
CREATE YOUR OWN →
PROMPT
0 / 2000
Describe the scene. Character handles identity. Hit Shuffle for a random starter prompt.
Output
Pick a character + prompt, hit Generate.
CHAT · LIVE

Talk to one of your characters

10 messages per credit · moderation included · API docs →
CHARACTER
SETTINGS
WHAT THESE DO →
PERSONALITY
sent as your system message
Pick a character
chat-v1 · streaming
Say hi. Every message is moderated; nothing is stored unless moderation flags it.
All content shown is computer-generated and purely fictional. No images depict any real individual, living or deceased. All characters portrayed are intended to be 18 or older. See our Content Moderation Policy.