Ask AI

Generate and Edit Images with Seedream Models

ByteDance's Seedream models offer state-of-the-art image generation capabilities, from text-to-image generation to advanced image-to-image editing. This guide demonstrates how to harness their full potential through AvalAI's unified API.

Introduction

Seedream models stand out with unique features:

Chain of Thought Reasoning (5.0): Intelligent prompt analysis and optimization
MJ-Style Aesthetics (5.0): Built-in support for Midjourney-style generation
Sequential Image Generation: Create batches of thematically related images
Multi-Image Blending: Combine multiple reference images with text prompts
High-Resolution Output: Generate images up to 4K resolution
Streaming Support: Real-time image generation with progressive updates
Advanced Control: Fine-tune generation with specialized parameters

Available Models

Model	Capabilities	Best For
`seedream-5-0-260128`	CoT reasoning, MJ-style aesthetics, Text-to-image, Image-to-image, Intelligent prompt optimization	Premium creative content, artistic generation, professional photography

Basic Usage

Text-to-Image Generation

Generate high-quality images from text descriptions:

python

from openai import OpenAI

client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

# Basic text-to-image generation
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="A majestic dragon soaring through clouds at sunset, digital art style, highly detailed, cinematic lighting",
    size="2K",
    response_format="url",
    extra_body={"sequential_image_generation": "disabled", "watermark": False},
)

print(f"Generated image: {response.data[0].url}")

Image-to-Image Editing

Transform existing images with text guidance:

python

# Image-to-image editing
with open("input_image.png", "rb") as image_file:
    response = client.images.edit(
        model="seedream-5-0-260128",
        image=image_file,
        prompt="Transform this landscape into a cyberpunk cityscape with neon lights and futuristic buildings",
        size="2K",
        response_format="url",
        extra_body={"sequential_image_generation": "disabled", "watermark": False},
    )

print(f"Edited image: {response.data[0].url}")

Advanced Features

Sequential Image Generation

Create batches of thematically related images automatically:

python

# Sequential generation - let the model decide how many images
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="A story of a magical forest through the four seasons, each showing different magical creatures and atmospheric changes",
    size="2K",
    extra_body={
        "sequential_image_generation": "auto",
        "sequential_image_generation_options": {"max_images": 6},
        "watermark": False,
    },
)

print(f"Generated {len(response.data)} images:")
for i, image in enumerate(response.data):
    print(f"Image {i+1}: {image.url}")

Multi-Image Blending

Combine multiple reference images to create something new:

python

# Multi-image blending
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="Create a modern architectural design that blends these different styles harmoniously",
    extra_body={
        "image": [
            "https://example.com/gothic-cathedral.jpg",
            "https://example.com/modern-skyscraper.jpg",
            "https://example.com/traditional-japanese-temple.jpg",
        ],
        "sequential_image_generation": "disabled",
        "size": "2K",
        "watermark": False,
    },
)

print(f"Blended architecture: {response.data[0].url}")

Streaming Generation

Get real-time updates as images are generated:

python

import requests
import json


def stream_image_generation():
    url = "https://api.avalai.ir/v1/images/generations"
    headers = {
        "Authorization": "Bearer your-avalai-api-key",
        "Content-Type": "application/json",
    }

    data = {
        "model": "seedream-5-0-260128",
        "prompt": "A series of concept art pieces for a fantasy video game, showing different environments and characters",
        "size": "2K",
        "stream": True,
        "sequential_image_generation": "auto",
        "sequential_image_generation_options": {"max_images": 5},
        "watermark": False,
    }

    response = requests.post(url, headers=headers, json=data, stream=True)

    for line in response.iter_lines():
        if line:
            try:
                chunk = json.loads(line.decode("utf-8").replace("data: ", ""))
                if "data" in chunk:
                    for image in chunk["data"]:
                        if "url" in image:
                            print(f"New image ready: {image['url']}")
                            print(f"Size: {image.get('size', 'Unknown')}")
            except json.JSONDecodeError:
                continue


# Run streaming generation
stream_image_generation()

Resolution and Quality Control

High-Resolution Generation

Generate images at different resolutions:

python

# 4K high-resolution generation
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="An ultra-detailed portrait of a cyberpunk character with intricate mechanical augmentations, professional photography lighting, 8K quality",
    size="4K",
    response_format="url",
    extra_body={"sequential_image_generation": "disabled", "watermark": False},
)

print(f"4K image: {response.data[0].url}")
print(f"Actual size: {response.data[0].size}")

Custom Aspect Ratios

Generate images with specific dimensions:

python

# Custom aspect ratio - wide cinematic
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="A cinematic wide shot of a spaceship battle in deep space, epic scale, dramatic lighting",
    size="3024x1296",  # 21:9 aspect ratio
    response_format="url",
    extra_body={"sequential_image_generation": "disabled", "watermark": False},
)

print(f"Cinematic image: {response.data[0].url}")

# Portrait orientation
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="A tall fantasy tower reaching into the clouds, vertical composition, detailed architecture",
    size="1440x2560",  # 9:16 aspect ratio
    response_format="url",
    extra_body={"sequential_image_generation": "disabled", "watermark": False},
)

print(f"Portrait image: {response.data[0].url}")

Creative Applications

Style Transfer and Artistic Effects

python

# Style transfer example
with open("input_image.png", "rb") as image_file:
    response = client.images.edit(
        model="seedream-5-0-260128",
        image=image_file,
        prompt="Transform this photo into a Van Gogh painting style with swirling brushstrokes and vibrant colors",
        size="2K",
        response_format="url",
        extra_body={"sequential_image_generation": "disabled", "watermark": False},
    )

print(f"Van Gogh style: {response.data[0].url}")

# Multiple style variations
with open("input_image.png", "rb") as image_file:
    response = client.images.edit(
        model="seedream-5-0-260128",
        image=image_file,
        prompt="Create artistic variations of this portrait in different styles: oil painting, watercolor, digital art, and pencil sketch",
        size="2K",
        extra_body={
            "sequential_image_generation": "auto",
            "sequential_image_generation_options": {"max_images": 4},
            "watermark": False,
        },
    )

print(f"Generated {len(response.data)} style variations:")
for i, image in enumerate(response.data):
    print(f"Style {i+1}: {image.url}")

Product Visualization

python

# Product mockup generation
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="A sleek modern smartphone on a minimalist desk setup, professional product photography, clean background, studio lighting",
    size="2K",
    response_format="url",
    extra_body={"sequential_image_generation": "disabled", "watermark": False},
)

print(f"Product mockup: {response.data[0].url}")

# Multiple product angles
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="Professional product shots of a luxury watch from multiple angles: front view, side profile, back view, and wrist shot",
    size="2K",
    extra_body={
        "sequential_image_generation": "auto",
        "sequential_image_generation_options": {"max_images": 4},
        "watermark": False,
    },
)

print(f"Generated {len(response.data)} product angles:")
for i, image in enumerate(response.data):
    print(f"Angle {i+1}: {image.url}")

Character Design and Concept Art

python

# Character design variations
response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="Fantasy character design: a powerful wizard with unique magical abilities, show different outfit variations and poses",
    size="2K",
    extra_body={
        "sequential_image_generation": "auto",
        "sequential_image_generation_options": {"max_images": 6},
        "watermark": False,
    },
)

print(f"Character variations: {len(response.data)} designs")
for i, image in enumerate(response.data):
    print(f"Design {i+1}: {image.url}")

JavaScript/Node.js Examples

Basic Usage with Node.js

javascript

import { OpenAI } from "openai";

const client = new OpenAI({
    apiKey: process.env.AVALAI_API_KEY,
    baseURL: "https://api.avalai.ir/v1",
});

async function generateImage() {
    try {
        const response = await client.images.generate({
            model: "seedream-5-0-260128",
            prompt: "A beautiful sunset over a mountain landscape, photorealistic, high detail",
            size: "2K",
            response_format: "url",
            // @ts-expect-error extra_body is a provider-specific parameter
            extra_body: {
                "sequential_image_generation": "disabled",
                "watermark": false
            }
        });

        console.log("Generated image:", response.data[0].url);
        return response.data[0].url;
    } catch (error) {
        console.error("Error generating image:", error);
    }
}

generateImage();

Sequential Generation with Node.js

javascript

async function generateImageSeries() {
    try {
        const response = await client.images.generate({
            model: "seedream-5-0-260128",
            prompt: "A time-lapse story of a city from dawn to midnight, showing the changing atmosphere and lighting",
            size: "2K",
            // @ts-expect-error extra_body is a provider-specific parameter
            extra_body: {
                "sequential_image_generation": "auto",
                "sequential_image_generation_options": {
                    "max_images": 8
                },
                "watermark": false
            }
        });

        console.log(`Generated ${response.data.length} images in the series:`);
        response.data.forEach((image, index) => {
            console.log(`Time ${index + 1}: ${image.url}`);
        });

        return response.data;
    } catch (error) {
        console.error("Error generating image series:", error);
    }
}

generateImageSeries();

Direct API Calls

Using cURL

bash

# Basic text-to-image generation
curl https://api.avalai.ir/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "seedream-5-0-260128",
    "prompt": "A futuristic city skyline at night with flying cars and neon lights",
    "sequential_image_generation": "disabled",
    "response_format": "url",
    "size": "2K",
    "watermark": false
  }'

# Sequential generation
curl https://api.avalai.ir/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "seedream-5-0-260128",
    "prompt": "A collection of fantasy creatures in their natural habitats",
    "sequential_image_generation": "auto",
    "sequential_image_generation_options": {
      "max_images": 5
    },
    "response_format": "url",
    "size": "2K",
    "watermark": false
  }'

# Image editing
curl https://api.avalai.ir/v1/images/edits \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "seedream-5-0-260128",
    "image": "https://example.com/input-image.jpg",
    "prompt": "Add magical elements and fantasy lighting to this scene",
    "sequential_image_generation": "disabled",
    "response_format": "url",
    "size": "2K",
    "watermark": false
  }'

Best Practices

Prompt Engineering

Be Specific and Detailed

python

# Good prompt
prompt = "A steampunk airship flying through storm clouds, brass and copper details, Victorian era design, dramatic lighting with lightning, highly detailed mechanical components, cinematic composition"

# Less effective prompt
prompt = "A cool airship"

Include Technical Specifications

python

prompt = "Professional product photography of a luxury watch, macro lens, studio lighting, white background, commercial quality, high resolution, sharp focus"

Specify Artistic Style

python

prompt = "Digital painting in the style of concept art, matte painting technique, photorealistic rendering, trending on ArtStation"

Performance Optimization

Choose Appropriate Resolution

python

# For web use
size = "1K"  # Faster generation

# For print or detailed work
size = "4K"  # Higher quality, slower generation

Use Streaming for Batch Generation

python

# Enable streaming for multiple images
extra_body = {"stream": True, "sequential_image_generation": "auto"}

Optimize Batch Sizes

python

# Balance between efficiency and resource usage
sequential_image_generation_options = {"max_images": 5}  # Sweet spot for most use cases

Error Handling

python

import time
from openai import OpenAI


def robust_image_generation(prompt, max_retries=3):
    client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

    for attempt in range(max_retries):
        try:
            response = client.images.generate(
                model="seedream-5-0-260128",
                prompt=prompt,
                size="2K",
                response_format="url",
                extra_body={
                    "sequential_image_generation": "disabled",
                    "watermark": False,
                },
            )

            # Check if generation was successful
            if response.data and response.data[0].url:
                return response.data[0].url
            else:
                raise Exception("No image URL in response")

        except Exception as e:
            print(f"Attempt {attempt + 1} failed: {str(e)}")
            if attempt < max_retries - 1:
                time.sleep(2**attempt)  # Exponential backoff
            else:
                raise e


# Usage
try:
    image_url = robust_image_generation("A beautiful landscape painting")
    print(f"Successfully generated: {image_url}")
except Exception as e:
    print(f"Failed to generate image after all retries: {e}")

Conclusion

Seedream 5.0 offers unprecedented capabilities for image generation and editing. Its unique features like sequential generation, multi-image blending, and streaming support make it ideal for both creative and commercial applications. By following the examples and best practices in this guide, you can harness the full potential of this state-of-the-art model through AvalAI's unified API.

Generate and Edit Images with Seedream Models ​

Introduction ​

Available Models ​

Basic Usage ​

Text-to-Image Generation ​

Image-to-Image Editing ​

Advanced Features ​

Sequential Image Generation ​

Multi-Image Blending ​

Streaming Generation ​

Resolution and Quality Control ​

High-Resolution Generation ​

Custom Aspect Ratios ​

Creative Applications ​

Style Transfer and Artistic Effects ​

Product Visualization ​

Character Design and Concept Art ​

JavaScript/Node.js Examples ​

Basic Usage with Node.js ​

Sequential Generation with Node.js ​

Direct API Calls ​

Using cURL ​

Best Practices ​

Prompt Engineering ​

Performance Optimization ​

Error Handling ​

Conclusion ​

Related Resources ​

Generate and Edit Images with Seedream Models

Introduction

Available Models

Basic Usage

Text-to-Image Generation

Image-to-Image Editing

Advanced Features

Sequential Image Generation

Multi-Image Blending

Streaming Generation

Resolution and Quality Control

High-Resolution Generation

Custom Aspect Ratios

Creative Applications

Style Transfer and Artistic Effects

Product Visualization

Character Design and Concept Art

JavaScript/Node.js Examples

Basic Usage with Node.js

Sequential Generation with Node.js

Direct API Calls

Using cURL

Best Practices

Prompt Engineering

Performance Optimization

Error Handling

Conclusion

Related Resources