Back to Models Guide
Bytedanceimage

Complete Guide to Using Seedream 5.0 Lite

Transform and edit images with ByteDance's multimodal Seedream engine at up to 3K resolution.

Try This ModelTutorial

Overview

Seedream 5.0 Lite Image-to-Image brings the same unified multimodal architecture to image editing and transformation. Upload one or more source images and describe the desired changes to produce edited outputs at up to 3K resolution.

The model understands both global and localized transformations. You can apply sweeping style changes across the entire image or describe specific edits to individual elements. It preserves structural integrity while making the requested modifications, resulting in natural-looking edits that blend seamlessly with the original content.

With multi-image input support, you can provide reference images for style matching or character consistency. Combined with the same quality tiers and aspect ratio options as the text-to-image variant, this model offers a complete image editing solution at a competitive 38.5 credits per generation.

Capabilities

  • Image editing and transformation via natural language instructions
  • Multi-image input for style and character reference
  • Output resolution up to 3K in High quality mode
  • Global style transfer and localized element editing
  • Eight aspect ratio options for flexible output framing
  • Structural preservation during transformations

Use Cases

1

Refining AI-generated images with targeted adjustments

2

Style transfer to match brand guidelines or visual themes

3

Scene modification and environment changes

4

Lighting and color grade adjustments on existing images

5

Creating variations of a base image for A/B testing

Input Parameters

prompt
textarearequired

A text description of the image you want to generate (max 3000 chars).

Describe the specific changes you want. Be explicit about what to modify and what to preserve. For example: 'Change the lighting to golden hour while keeping the subject and pose identical'. Maximum 2996 characters.

image_urls
filerequired

Upload image file(s) to use as input (JPEG/PNG/WebP, ≤10MB each, up to 14).

Upload one or more source images. The primary image is the one to edit. Additional images can serve as style or character references.

aspect_ratio
selectrequired

Width-height ratio of the image, determining its visual form.

Choose an aspect ratio that matches or is close to your source image to avoid unwanted cropping. If changing the ratio intentionally, the model will adapt the composition.

Options
1:14:33:416:99:162:33:221:9
Default: 1:1
quality
selectrequired

Basic outputs 2K images, while High outputs 4K images.

Basic (2K) for fast iteration and previews. High (3K) for final production output where detail matters.

Options
BasicHigh
Default: basic
NSFW Filter
toggle

Enable NSFW content filtering.

Default: false

Tips & Best Practices

Describe changes, not the final result
Use multiple reference images for consistency
Preserve aspect ratio for clean edits
Chain edits for complex transformations

Related Models

Seedream 5.0 Lite

View Guide →

Seedream 4.5 Edit

View Guide →

Seedream 4.0 Edit

View Guide →