Codia Image Generate

codia-image-generate

Runtime And Auth

Use the local codia-design CLI for this feature. Before calling the API, verify that codia-design is available. If it is missing and global npm installs are allowed, install it with npm install -g @codia-ai/codia-design-cli. Then check codia-design auth status; if it is not connected, run codia-design auth login --platform codex or bind an API key with codia-design auth set --api-key api_key_xxx. Never print API keys or ~/.codia/design-skills/config.json.

Generate images based on text descriptions. Supports multiple models, each with different styles and credits consumption.

CLI Command

codia-design image generate \
  --prompt "TEXT" \
  [--size WxH] \
  [--n COUNT] \
  [--model MODEL] \
  [--quality low|medium|high] \
  [--background transparent|opaque] \
  [--reference_images '["URL1","URL2"]'] \
  [--download-dir DIR] \
  [--no-download] \
  [--out FILE]

Parameters

Credit Matrix

Prices are charged from the unified Codia Open API balance. The CLI estimates from open_api_v2; use codia-design limits or https://codia.ai/api-reference#description/introduction for the live pricing table.

Text-to-image and image-to-image

image generate charges per generated image and multiplies by --n. image image-to-image charges per request. Both use this model table.

GPT Image uses quality-specific pricing:

Remix

Reframe

Upscale

Model Guidance

CLI Defaults

When the user simply asks to "generate an image" or describes a style but does not specify a model/size, the CLI proactively provides easy-to-use defaults:

• Default model: seedream_5.
• Default size: 2560x1440, the 2K supported size for Seedream 16:9.
• When the user passes in --model or --size, the user parameters shall prevail.
• Download data.image_urls to the current local directory by default, and append data.local_file / data.local_files to the response; use --download-dir to specify the directory, or use --no-download to turn it off.
• When the user requires precise output of pixels, it means that the image generation model may be normalized according to server-side rules; use post-processing resize/crop when a fixed final size is required.

Size Normalization

--size is the requested size, which is not guaranteed to be equal to the final image size. The server will perform normalization according to model constraints:

• The default model and most 1024 series models use the 1024 base size.
• seedream_5 / seedream_4_5 requires the minimum area to be no less than 1920x1920 and the single side no more than 4096. For example, if the requested area of 1920x1080 is insufficient, it will be normalized to 2560x1440 according to 16:9.
• ideogram_v3 uses its own native size table.

Commonly used ratio references:

Response

{
  "ok": true,
  "data": {
    "image_urls": [
      "https://cdn.codia.ai/generated/abc123.png"
    ]
  }
}

Usage Examples

# Basic generation
codia-design image generate --prompt "The sea after the rain, realistic photography style" --download-dir ./outputs

# Only save JSON, do not download images
codia-design image generate --prompt "minimalist app icon for a design tool" --no-download --out result.json

# Transparent background icon (gpt_image)
codia-design image generate \
  --prompt "3D app icon with geometric shapes" \
  --model gpt_image \
  --quality high \
  --background transparent \
  --out icon.json

# Generation of reference images
codia-design image generate \
  --prompt "product photo on marble table" \
  --model nano_banana_pro \
  --reference_images '["https://example.com/style1.jpg"]' \
  --out result.json

# Generate 4 photos in batches
codia-design image generate --prompt "hero banner" --n 4 --out batch.json

Read the result URL:

const { data } = require('./result.json');
data.image_urls.forEach((url, i) => console.log('Image', i, url));

Errors & Billing

Credits: See model table credits/image × n. The CLI prints an open_api_v2 estimate before running. After the command completes, it checks available_credits and reminds the user if the remaining balance is below 2x the estimated cost; this does not block generation. Use --estimate to show cost without sending the request. Extra reference images can add surcharges for some models. · Duration: Synchronous, usually 5–20s