analyze

Analysis

Analyze Images

Run analysis tools on images or text. Actions include caption, describe_style, detect, embed, prompt_gen, and translate. For local files, upload them first with upload_asset and pass the resulting asset_id. Args: - action: required analysis mode - image: asset_id or URL for image-based actions - text: required for embed and translate, optional for prompt_gen - num_results, model_id, parameters, wait: optional action-specific controls - team_id, project_id: required for OAuth callers Returns: action-specific analysis output. prompt_gen returns prompt suggestions synchronously; the other actions may return immediate results or a polled job result depending on the backend. Examples: - "Caption this image" -> action="caption", image="asset_xxx" - "Generate a depth map" -> action="detect", image="asset_xxx", parameters={"modality": "depth"} - "Translate this prompt" -> action="translate", text="..." Don't use when: You need to generate final user-facing assets. Prefer run_model instead.

read-onlyidempotentopen-world

Parameters

Name	Type	Required	Description
action	enum(caption \| describe_style \| detect \| embed \| prompt_gen \| translate)	✓	Analysis action: caption (describe image), describe_style (analyze style), detect (edge/depth/pose maps), embed (vector embedding), prompt_gen (generate/enhance prompts), translate (translate prompts)
image	string	—	Source image URL or asset ID.
text	string	—	Text input (required for embed/translate, optional for prompt_gen).
num_results	number	1	Number of result variations (prompt_gen only).
model_id	string	—	Analysis model ID (optional).
parameters	record	—	Action-specific params (e.g., modality for detect, language for translate).
wait	boolean	true	Wait for completion (up to 5 min).
team_id	string	—	Team ID. Required if user belongs to multiple teams.
project_id	string	—	Project ID to scope the operation to.
response_format	enum(json \| markdown)	json	Output format: 'json' for structured data, 'markdown' for human-readable text.

Example Request

JSON

{
  "action": "caption",
  "image": "asset_img001",
  "team_id": "team_abc123",
  "project_id": "proj_xyz789"
}

Example Response

JSON

{
  "action": "caption",
  "status": "succeeded",
  "caption": "A majestic dragon with iridescent scales perched on a rocky mountain peak at golden hour, dramatic clouds in the background."
}

Common Use Cases

Auto-caption game assets for searchability and accessibility
Analyze image style with describe_style before fine-tuning a model on similar content
Generate enhanced prompts from rough descriptions using prompt_gen
Extract depth or edge maps with detect to use as control inputs for other models