Mira Image

Mira Image is our image generator and editor. Two modes share one endpoint: text-to-image (classic prompt-based generation) and image-to-image (edit an existing image or compose multiple reference inputs).

Capabilities

• Text-to-image — generate from a text prompt in 50+ languages
• Image-to-image — edit or extend an existing image passed as a reference input
• Multi-reference — up to 4 reference images per request — transfer style, identity, product, background
• Resolutions — square (1024×1024), portrait (1024×1536) and landscape (1536×1024)
• Quality — low / medium / high / auto — pick the speed-vs-detail balance
• Content moderation — all requests pass a safety check before generation; disallowed prompts return 400

Text-to-image

curl https://api.vmira.ai/v1/images/generate \
  -H "Authorization: Bearer $MIRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A quiet lighthouse on a rocky shore at sunset, cinematic light",
    "size": "1024x1024",
    "n": 1,
    "quality": "high"
  }'

Python (OpenAI-compatible SDK)

from openai import OpenAI

client = OpenAI(
    api_key="sk-mira-YOUR_API_KEY",
    base_url="https://api.vmira.ai/v1",
)

result = client.images.generate(
    prompt="Minimalist coffee-shop logo, flat vector",
    size="1024x1024",
    n=1,
)

print(result.data[0].url)

Image-to-image (with references)

Pass one or more image URLs or data-URIs in reference_images. The prompt describes what should change or how to combine the references.

curl https://api.vmira.ai/v1/images/generate \
  -H "Authorization: Bearer $MIRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Same product, but on a marble countertop with soft morning light",
    "size": "1024x1024",
    "n": 1,
    "reference_images": [
      "https://example.com/product.png"
    ]
  }'

Response shape

{
  "created": 1714000000,
  "data": [
    {
      "url": "https://cdn.vmira.ai/img/abc123.png",
      "revised_prompt": "Expanded prompt actually used for generation"
    }
  ]
}

Parameters

• prompt — required, 1–4000 characters
• size — 1024x1024 | 1024x1536 | 1536x1024 | auto
• n — number of images (1–4)
• quality — low | medium | high | auto
• reference_images — array of URLs or data-URIs, up to 4 — enables image-to-image mode