API Documentation

Check your team's credit balance, plan status, and concurrent request limit. Call this before generation requests.

# Response
{
  "credits": 99999,
  "plan_status": "free",
  "concurrent_requests": 10,
  "plan_name": null,
  "subscription_id": null
}

Get paginated credit usage history for a program.

# Response
{
  "items": [
    { "program_id": "xxx", "request_id": "xxx", "credits": 1, "timestamp": 1773702421471, "operation": "model_gen" }
  ],
  "pagination_token": null
}

List programs accessible to your team.

# Response
{ "programs": [{ "program_id": "xxx", "program_name": "My Brand", "create_time": 1773702421471 }], "pagination_token": null }

Create a new program (workspace) within your team.

# Response
{
  "program_name": "My New Brand",
  "user_id": "xxx", "create_time": 1773702421471
}

List projects, or get a single project detail by adding project_id.

# List response (no project_id)
{ "projects": [{ "project_id": "xxx", "project_name": "Spring", "project_status": "Created" }], "pagination_token": null }

# Detail response (with project_id)
{ "project_id": "xxx", "project_name": "Spring", "project_status": "Created", "create_time": 1773702421471 }

Create a new project within a program.

Upload garment images (base64) or initiate multipart zip upload for SKU archives.

# For garment_image → { "success": true, "image_id": "xxx" }
# For sku_zip → { "message": "success", "key": "...", "upload_id": "...", "upload_urls": ["..."] }

Complete a multipart zip upload after all parts are uploaded to presigned URLs.

# Response: { "message": "success" }

Create a new AI model in the gallery with searchable attributes. Add images after with /model/image.

# Request with attributes
{ "model_name": "Caucasian Plus-Size Woman",
  "model_type": "Women", "model_size": "Plus size",
  "attributes": { "ethnicity": "Caucasian", "age_range": "Middle-aged",
    "hair_color": "Blonde", "hair_style": "Long", "skin_tone": "Fair" } }
# Response: {"success": true, "model_id": "xxx"}

List models with attribute filters, or get a model's images by adding model_id. LLM agents: use filters to match user descriptions.

# List response — with attributes for LLM selection
{ "models": [{ "model_id": "xxx", "model_name": "Dae Nakamura", "model_type": "Women",
  "attributes": { "ethnicity": "East Asian", "age_range": "Young Adult",
    "hair_color": "Black", "skin_tone": "Light" } }] }

# Images response (with model_id)
{ "model_images": [
  { "model_image_id": "...", "model_image_url": "https://...",
    "model_id": "xxx", "style_id": "yyy",
    "view_type": "front_facing", "crop_type": "fullbody" }
] }
# view_type: front_facing, back_facing (or null)
# crop_type: fullbody, head_half_cropped, top, bottom (or null)

Generate AI face images. Supports variety (diverse) and fidelity (precise) modes.

# Response
{ "request_id": "xxx", "status": "Request Submitted", "images_generated": [], "num_output_images": 2 }

# Poll via GET /model/image?request_id=xxx until status = "Generation Completed"
# Then save to model via POST /model/image with model_id + image_url from result

Prompt template: Headshot front facing portrait of a {Age} {Race} {BodySize} {Gender} with {HairColor} {HairStyle} hair with {EyeColor} eye with {SkinColor} skin tone in the {Background} setting with {Expression}. {Pose} pose, professional lighting, high detail.

Two modes: generate a new image (with prompt) or upload/save an existing image to a model (with model_id).

Mode 1 — Generate (provide prompt):

# Generate request
{ "prompt": "professional photo of model wearing garment" }

# Generate response — poll GET /model/image?request_id=xxx for results
{ "request_id": "xxx", "status": "Request Submitted", "images_generated": [], "num_output_images": 1 }

Mode 2 — Upload/Save to model (provide model_id):

# Upload request — save a generated image to a model
{ "model_id": "xxx", "style_id": "yyy", "image_url": "https://s3.../generated.png?signed..." }

# Upload response
{ "success": true, "image_id": "...", "image_url": "https://...", "view_type": "front_facing", "crop_type": "fullbody" }
# view_type/crop_type are auto-detected by AI: front_facing/back_facing, fullbody/head_half_cropped

Poll generation status for POST /model/face, POST /model/image (generate), or POST /model/pose.

# Response (completed)
{
  "status": "Generation Completed",
  "images_generated": [
    { "model_image_id": "path/to/output.png", "model_image_url": "https://s3.../output.png?signed..." }
  ]
}

Generate model images in specific poses using a reference image. Costs credits.

# Response
{ "request_id": "xxx" }

# Poll: GET /model/image?request_id=xxx → same polling as face/image generation
# Save results: POST /model/image with model_id + style_id + image_url from result
# The reference_image's model_id and style_id can be found via GET /model?model_id=

Browse 143 preset poses. Filter by model_category and prompt_type. Use pose reference images with POST /model/pose.

# Response
{ "poses": [{ "pose_id": "f00cb596...", "pose_name": "Full body upright",
  "model_category": "Women", "prompt_type": "fullbody",
  "reference_image_url": "https://cdn.huhu.ai/poses/xxx.png",
  "sketch_url": "https://cdn.huhu.ai/poses/yyy.png" }], "total": 14 }

# LLM agent tips: "front view" → fullbody + pose_name with "front"
# "back view" → pose_name with "back", "close-up" → close_up_upper_body
# Use reference_image_url as reference_image in POST /model/pose

List your custom poses, or get detail by adding pose_id. For preset poses, use GET /pose/library.

# List response
{ "poses": [{ "pose_id": "xxx", "title": "Standing Front", "prompt": "standing pose, front facing...",
  "reference_image_presigned_url": "https://...", "prompt_type": "fullbody" }] }

# Detail response (with pose_id)
{ "pose_id": "xxx", "title": "...", "prompt": "...", "reference_image_presigned_url": "https://..." }

Create a new pose from a reference image. AI extracts the pose prompt automatically.

# Response
{ "status": "success", "pose_id": "xxx", "pose_prompt": "standing pose, front facing, arms at sides...", "pose_title": "My Pose" }

# Use the pose_prompt in POST /model/pose to generate images in this pose

List SKUs in a project, or get detail by adding sku_id. Detail includes AI-classified garment images.

# Detail response (with sku_id) — includes AI garment classification
{
  "skus": [{
    "sku_id": "xxx", "sku_name": "sku1", "garment_type": "Full body",
    "garment_images": [
      { "garment_id": "abc", "image_id": "assets/sku_images/xxx.png", "image_url": "https://...",
        "garment_type": "Top", "view": "front", "presentation": "flatlay_clean" },
      { "garment_id": "def", "image_id": "assets/sku_images/yyy.png", "image_url": "https://...",
        "garment_type": "Bottom", "view": "front", "presentation": "flatlay_clean" },
      { "garment_id": "ghi", "image_id": "assets/sku_images/zzz.png", "image_url": "https://...",
        "garment_type": "Shoes", "view": "front", "presentation": "flatlay_clean" }
    ]
  }]
}

Garment classification: After zip upload, images are automatically classified by AI into garment_type (Top, Bottom, Shoes, Accessory) and view (front, back). This classification is used when creating tasks.

Update SKUs — assign model, style, garment type.

# Body: { "sku_ids": ["sku-001"], "model_id": "...", "style_id": "...", "garment_type": "Full body" }

Create generation tasks for SKUs. Tasks are auto-created based on garment image classification (front/back views).

# Response: {"message": "success"}
# Tasks are created based on garment pairing (front view task + back view task)
# Then GET /task to find new task IDs with status "Task created"

Update a task — set model_image before generation. REQUIRED for try-on.

Submit try-on generation. Each task must have model_image set first.

# Response: {"message": "success"}
# Errors: 429 (concurrent limit exceeded)
# Poll: GET /task → ~90s to complete

List tasks (poll status), or get task detail with output images by adding project_id + sku_id + task_id.

# List response (no detail params) — poll for generation progress
{ "skus": [{ "sku_name": "Blue Dress", "tasks": [{ "task_id": "xxx", "task_status": "Generation completed" }] }] }

# Detail response (with project_id + sku_id + task_id) — get output images
{ "task_id": "xxx", "task_status": "Generation completed", "output": [{ "image_id": "xxx", "image_url": "https://..." }] }

Generate a video from a try-on output image. Costs 25 credits (5s) or 40 credits (10s).

Poll video generation status. Returns video_url when completed.

# Response
{
  "requests": [{
    "request_status": "Generation completed",
    "video_url": "https://s3.../video.mp4?signed...",
    "duration": "5"
  }]
}

Upscale an image to 2x-4x resolution (max 18M output pixels). Costs 2 credits per image.

# Response
{ "status": "success", "request_id": "req-upscale-123", "cost": 2 }

# Resolution limits: input_pixels × factor² ≤ 18,000,000
# 1024×1024 → 4x = 16.8M ✓   |   2048×2048 → 4x = 67M ✗

Poll upscale status. Returns result URLs when completed.

# Response (completed)
{ "request_id": "...", "status": "success", "result_urls": ["https://...upscaled.png"] }