AI Fashion Generation API for Agents: Product Photos, Try-On Images, Edits, and Video

If you ask a coding agent like ChatGPT, Codex, Claude, or Cursor to build a fashion visual generation workflow, the hardest part is rarely writing the HTTP request. The hard part is choosing models, understanding fashion-specific inputs, waiting for async jobs, handling result URLs, and turning one successful generation into a repeatable product workflow.

Uwear rebuilt its API around that reality. The API is open, documented for agents, and built around a simple async generation contract that can produce on-model product photos, virtual try-on style images, edits, upscales, and short fashion videos from one consistent workflow.

The Agent Problem With Generic Image APIs

Generic image generation APIs are powerful, but fashion workflows have extra constraints. A dress is not just a prompt. It has a front view, sometimes a back view, fabric behavior, print fidelity, sizing cues, brand styling, camera angle, model consistency, and often a follow-up workflow: edit the background, upscale the result, animate it, save it as an avatar, or use it in a catalog automation.

That is why a useful agent-facing fashion API needs more than "send prompt, get image". It needs stable endpoints, explicit model choices, a predictable async lifecycle, result retrieval, and instructions that an LLM can load without reverse-engineering your product.

What the Uwear API Gives an Agent

Uwear exposes a REST API for fashion visual generation. An agent can use it to manage clothing items, create generation jobs, poll job status, receive webhooks, retrieve generated results, save avatars, edit images, upscale outputs, and turn static fashion images into videos.

The most important design change is that image generation, edits, upscaling, and video generation now fit into the same mental model: create a generation job, wait for a terminal state, then download the resulting assets.

Give Your Agent the UWEAR.md Skill

Agents work best when they have a compact, task-specific reference instead of a vague landing page. That is why Uwear now has a downloadable UWEAR.md reference for LLMs and coding agents.

# Agent setup

1. Download https://uwear.ai/UWEAR.md
2. Give it to ChatGPT, Codex, Claude, Cursor, or your coding agent.
3. Add your Uwear API key as a secret.
4. Ask the agent to create a fashion image/video generation workflow.
5. Use polling for simple scripts or webhooks for production automation.

The file tells the agent which routes to use, how Uwear names fields, how generation status works, where results appear in polling versus webhook payloads, which model slugs are available, how to structure bulk collection work, how to plan a platform try-on flow, and how long generated asset URLs remain available.

One API, Multiple Fashion Visual Workflows

Uwear is not only a virtual try-on API, and it is not only an AI product photography API. It is a fashion visual generation API with several related workflows that share the same generation surface.

Curated Image and Video Models for Fashion

Agents can call generic image models directly, but then the agent also has to decide which model works for which garment, whether the model supports references, whether it supports a target aspect ratio, and whether it can produce video or upscale. Uwear puts a curated fashion model catalog behind a single API surface.

That gives an agent access to many image and video models through one fashion-specific API, instead of forcing the user to wire every provider separately and rediscover which settings work for apparel.

Example: Generate On-Model Fashion Product Photos

A simple agent workflow starts by creating or referencing a clothing item, then creating a generation job. The agent can choose a virtual model automatically or use a saved `avatar_id` for consistency.

POST /generation
Authorization: Bearer YOUR_API_KEY

{
  "clothing_item_id": 123,
  "model_slug": "nano-banana-2",
  "use_case": "generate",
  "prompt": "female model wearing the clothing, professional studio photography",
  "num_images": 2,
  "camera": "full_body_front",
  "aspect_ratio": "9:16",
  "resolution": "2K",
  "enhance_user_prompt": true,
  "avatar_id": null
}

The response returns a `generation_id`. For a small script, the agent can poll. For a production workflow, it can use webhooks.

GET /generation/789
Authorization: Bearer YOUR_API_KEY

// Poll until status is Done or Error.
// Polling responses put generated assets in generation_results.

Polling for Scripts, Webhooks for Agents in Production

Uwear supports both polling and webhooks because agents operate in different contexts. A local Codex script might poll until a generation finishes. A production catalog pipeline should usually register a webhook and let Uwear call back when the job reaches a terminal state.

Webhooks can be configured once on the API Keys page, or sent per request with `webhook_url`, `webhook_secret`, and `webhook_events`. Signed payloads, delivery history, testing, retries, and replay support make this usable for real automation instead of demos only.

Example: Turn a Generated Image Into Fashion Video

Once an agent has a generated image, it can reuse the result as an input for video. The same `/generation` endpoint handles the next step.

POST /generation
Authorization: Bearer YOUR_API_KEY

{
  "generation_result_parent_id": 456,
  "model_slug": "kling-3-pro",
  "use_case": "video",
  "num_images": 1,
  "prompt": "fashion model posing for a clean ecommerce product video",
  "duration": 5,
  "generate_audio": false
}

This is where Uwear becomes especially useful for agents. A user can ask for a complete fashion asset workflow, not a single model call: generate product photos, pick the best result, upscale it, create a short video, then download every output before temporary URLs expire.

What an Agent Can Build With Uwear

Why Uwear Is a Good Option for Agentic Fashion Generation

Uwear is a good fit when the user wants fashion visuals rather than general-purpose image generation. The API has the model breadth agents need, but also the domain structure they otherwise have to invent: clothing items, avatars, camera semantics, use cases, image results, video results, credits, polling, and webhooks.

Start With the Skill File

The fastest way to test Uwear with an agent is to download the reference file, give it to the agent, and ask it to build a small generation script using your API key.