What it does

image-generate-localize composes two existing endpoints behind one paid call: image-generate and translate. Send a prompt and a target_language, and the response comes back with the generated image URL and a caption in both the original text and the localized translation. 

{
  "prompt": "A neon-lit night market street food stall, cinematic lighting",
  "target_language": "French"
}

returns 

{
  "image_url": "https://fal.media/files/.../generated.png",
  "caption": {
    "original": "A neon-lit night market street food stall, cinematic lighting",
    "localized": "Un stand de rue de marche nocturne neon, eclairage cinematographique",
    "language": "French"
  },
  "composed_of": ["image-generate", "translate"],
  "degraded": false
}

Pass an optional caption field when the text that goes with the image should differ from the generation prompt — a social post caption, for example, rather than the raw prompt.

Why this pair

Image generation and translation are the most common pairing we see agents stitch together back-to-back: a caller generates an image, then translates the caption or post text that goes with it for a non-English audience. image-generate-localize collapses that into one call instead of two round trips.

Degradation

Image generation is required — if it fails, the call fails. Translation is best-effort: if the translate leg times out or errors, the response still returns the generated image with the caption's localized field falling back to the original text, and the call is marked degraded.

Priced at $0.04, in the compose cluster alongside image-caption-localize and weather-localize-pack.