AI Image Generator — Ai Generator
Generate images from text prompts using state-of-the-art AI models like Flux, Stable Diffusion, and Midjourney.
What This Tool Does
The AI Image Generator creates images from text descriptions using state-of-the-art AI models. Provide a detailed text prompt and receive a high-quality image URL in return. Supports multiple AI models, custom dimensions, negative prompts, and reproducible seeds.
Available Models
| Model | Speed | Quality | Best For |
|---|---|---|---|
| flux-schnell | Fast (~5–10s) | Good | Rapid prototyping, iteration, batch generation |
| flux-dev | Medium (~15–30s) | High | Production assets, detailed scenes, final renders |
| stable-diffusion | Medium (~15–20s) | Good | Artistic styles, illustrations, creative work |
Default: flux-schnell — the best balance of speed and quality for most use cases.
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
| prompt | string | Yes | — | Text description of the image to generate |
| model | string | No | flux-schnell | AI model: flux-schnell, flux-dev, stable-diffusion |
| width | integer | No | 1024 | Image width in pixels (256–2048) |
| height | integer | No | 1024 | Image height in pixels (256–2048) |
| num_images | integer | No | 1 | Number of images to generate (1–4) |
| negative_prompt | string | No | — | What to exclude from the image |
| seed | integer | No | — | Fixed seed for reproducible results |
Prompt Writing Tips
Writing a good prompt is the most important factor in getting high-quality results.
Be specific about the subject:
❌ "a dog" → ✅ "a golden retriever puppy sitting on a wooden porch, looking at the camera"
Add style and medium:
"product photo, studio lighting, white background, 4K, photorealistic" "digital painting, vibrant colors, fantasy art, detailed" "minimalist illustration, flat design, pastel colors"
Specify the mood and lighting:
"soft morning light", "dramatic shadows", "warm sunset tones", "bright and airy"
Describe composition:
"close-up portrait", "wide-angle landscape", "bird's eye view", "rule of thirds"
Use negative prompts to remove unwanted elements:
"blurry, distorted, watermark, text, signature, low quality, ugly, deformed"
Example of a well-structured prompt:
A modern MacBook laptop on a clean wooden desk, minimalist home office setup,
soft natural window light from the left, shallow depth of field, product photography,
4K, photorealistic
Use Cases
- Marketing visuals — Generate on-brand images for ads, social media, and landing pages without a photo shoot.
- Product mockups — Visualise product concepts or packaging designs before production.
- Content creation — Produce blog header images, thumbnails, and article illustrations at scale.
- Design prototyping — Quickly explore visual directions and styles before committing to a full design.
- Social media — Create unique, eye-catching images for posts, stories, and banners.
Example Request
{
"prompt": "A modern laptop on a wooden desk, minimalist style, soft lighting, product photography",
"model": "flux-schnell",
"width": 1024,
"height": 1024
}
Example Response
{
"imageUrl": "https://storage.mapiok.com/results/abc123/image.png",
"model": "flux-schnell",
"prompt": "A modern laptop on a wooden desk, minimalist style, soft lighting, product photography",
"width": 1024,
"height": 1024,
"seed": 1749302,
"generationTimeMs": 7340
}
Limits and Tips
- Resolution: Width and height must be between 256 and 2048 pixels. Very large images (>1536px per side) increase generation time significantly.
- Aspect ratios: Common ratios work best — 1:1 (1024×1024), 16:9 (1024×576), 9:16 (576×1024), 4:3 (1024×768).
- Reproducibility: Set a
seedvalue to generate the same image repeatedly. The same prompt + seed + model always produces the same result. Omit seed for random variation. - Negative prompts: Keep negative prompts short and specific. Long negative prompt lists can reduce output diversity. Focus on 3–5 key exclusions.
- Credits: Each generation costs 5 credits regardless of model or size.
- Processing time:
flux-schnelltypically completes in 5–10 seconds; other models may take 15–30 seconds.
On this page