# Seedream 4.5 - Core.Today AI API

> Upgraded ByteDance image model with stronger spatial understanding and world knowledge. Supports single/multi-reference image-to-image editing and sequential (multi-image) generation.

- **Provider**: ByteDance
- **Model ID**: bytedance/seedream-4.5
- **Category**: Image Generation
- **Credits**: 90 per image — sequential mode (auto) scales per generated image, up to max_images (15)
- **Speed**: Medium
- **Quality**: Ultra

## Features
- Strong spatial understanding and world knowledge
- Multi-reference image-to-image generation (up to 14 input images)
- Sequential generation mode producing up to 15 related images in one call
- 2K and 4K resolution, plus custom width/height
- Wide range of aspect ratios including auto-match to input image

## Use Cases
- Story scene sets and character variation sheets in one call
- Multi-reference product photography compositing
- Style transfer and photo editing guided by reference images
- High-resolution (4K) marketing and print imagery

## API Endpoint

Base URL: https://api.core.today/v1
Create Prediction: POST /predictions
Get Status: GET /predictions/{job_id}

## Authentication
Header: X-API-Key: YOUR_API_KEY

## Input Parameters

### Required
- **prompt**: string - Text prompt for image generation (max 4000 characters)

### Optional
- **image_input**: array (default: []) - Input image(s) for image-to-image generation (1-14 images for single/multi-reference)
- **size**: string (default: 2K) - Image resolution: 2K, 4K, or custom (1K is not supported)
  Options: 2K, 4K, custom
- **width**: integer (default: 2048) - Custom image width, 1024-4096 (only used when size=custom)
- **height**: integer (default: 2048) - Custom image height, 1024-4096 (only used when size=custom)
- **aspect_ratio**: string (default: match_input_image) - Image aspect ratio. Ignored when size=custom
  Options: match_input_image, 1:1, 4:3, 3:4, 4:5, 5:4, 16:9, 9:16, 3:2, 2:3, 21:9, 9:21
- **sequential_image_generation**: string (default: disabled) - 'disabled' generates one image; 'auto' lets the model generate multiple related images
  Options: disabled, auto
- **max_images**: integer (default: 1) - Max images to generate when sequential_image_generation=auto (1-15). Billed per generated image
- **disable_safety_checker**: boolean (default: false) - Disable the safety checker for generated images

## Examples

### 4K Interior Photography (Default)
High-resolution single-image generation at 4K
```json
{
  "model": "bytedance/seedream-4.5",
  "input": {
    "prompt": "A warm, nostalgic film-style interior of a cozy café, shot on 35mm-inspired digital photography with soft afternoon sunlight filtering through the front windows.",
    "size": "4K",
    "aspect_ratio": "16:9",
    "sequential_image_generation": "disabled"
  }
}
```

### Sequential Story Scenes
Generate up to 15 related images (e.g. story scenes or character variations) in one call
```json
{
  "model": "bytedance/seedream-4.5",
  "input": {
    "prompt": "A children's book spread showing a fox exploring a forest across four consecutive scenes",
    "sequential_image_generation": "auto",
    "max_images": 4
  }
}
```

## Response Format
```json
{
  "job_id": "abc123",
  "status": "pending | processing | completed | failed",
  "result": "URL or data (when completed)"
}
```

## Usage Flow
1. POST /predictions with model and input -> receive job_id
2. GET /predictions/{job_id} -> poll until status is completed or failed
3. Result contains output URL(s)

## Tips
- sequential_image_generation=auto bills per generated image (up to max_images) — set max_images conservatively to control cost
- Use image_input with up to 14 reference images for multi-reference composition and character consistency
- 1K resolution is not supported — use 2K (default) or 4K for higher fidelity
- aspect_ratio is ignored when size=custom; set width/height instead

## Documentation
https://replicate.com/bytedance/seedream-4.5