> ## Documentation Index
> Fetch the complete documentation index at: https://docs.elumenta.ru/llms.txt
> Use this file to discover all available pages before exploring further.

# Image Generation Guide

> Best practices for prompt writing and model selection

## Choosing the Right Model

| Use Case                   | Recommended Model                     |
| -------------------------- | ------------------------------------- |
| Photorealistic photos      | `flux-1.1-pro` or `gpt-image-1.5`     |
| Product shots              | `flux-1.1-pro`                        |
| Artistic / illustration    | `midjourney-v6` or `midjourney-turbo` |
| Quick generation           | `dall-e-3` (5–13 tkn)                 |
| High-res 2K output         | `nano-banana-pro`                     |
| Image editing / inpainting | `flux-kontext-pro`                    |

## Prompt Writing Tips

**Be specific:** Instead of `"a car"`, write `"a red 1967 Ford Mustang on Route 66 at sunset, wide angle, film grain"`

**Use style references:** `cinematic`, `studio photography`, `oil painting`, `8k`, `photorealistic`, `bokeh`

**Negative prompts** (Flux, SD): Use them to remove common artifacts: `"blurry, low quality, distorted hands, watermark, text"`

## Error Handling

If image generation fails, the system will display an error message for 6 seconds. Common error types include:

* **Model overloaded**: The model is temporarily under high demand. Wait a few minutes or try a different model.
* **Generation timeout**: The request took too long to complete. Try again with a simpler prompt.
* **Content policy violation**: Your prompt was blocked by moderation filters. Modify your description.
* **Insufficient tokens**: Top up your account balance.
* **Network issues**: Check your internet connection.

## Aspect Ratio Guide

| Use Case           | Size                 |
| ------------------ | -------------------- |
| Social square      | `1024x1024`          |
| Portrait / mobile  | `768x1344` or `9:16` |
| Landscape / banner | `1344x768` or `16:9` |
| Story              | `1024x1536`          |

## Reproducible Results

Use the `seed` parameter to get the same image with the same prompt:

```python theme={null}
result = requests.post(
    "https://elumenta.ru/api/v2/generate",
    headers={"Authorization": "Bearer nb_YOUR_API_KEY"},
    json={
        "model": "flux-1.1-pro",
        "prompt": "Blue gradient abstract background",
        "seed": 12345
    }
)
# Same seed + same prompt = very similar output
```

## Token Costs

| Model              | Tokens |
| ------------------ | ------ |
| `photon-flash`     | 1      |
| `dall-e-3`         | 5–13   |
| `flux-1.1-pro`     | 7      |
| `flux-kontext-pro` | 7      |
| `midjourney-v6`    | 14     |
| `midjourney-turbo` | 28     |
| `gpt-image-1.5`    | 1–21   |
| `sd-3.5-large`     | 12     |
| `recraft-v3`       | 7      |
| `photon-1`         | 3      |
| `nano-banana-pro`  | 16     |

<Warning>
  DALL-E 3 and GPT Image 1.5 charge more for non-square aspect ratios (up to 2×). Use [Calculate Price](/en/api-reference/models/calculate-price) to check exact cost.
</Warning>