Seedream 4.0 (by Bytedance)

Overview

Consistent, reasoning-driven text→image model with high-resolution outputs.


Quick facts

  • Modes: Text → Image · Image → Image (editing)

  • Outputs / sizes: 1024×1024 (1:1), 1280×720 (16:9), 720×1280 (9:16) — higher when selecting 2K/4K

  • Aspect ratios: 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21


What it’s great for

  • High-detail photoreal images and editorial art.

  • High-resolution asset prototyping.

  • Image editing and inpainting tasks.


Example outputs

circle-info

Shadow study — experimental fashion film still

circle-info

Cinematic character moment

circle-info

Detailed environment concept

Copy-and-paste prompts


Parameters (editor-ready)

Name
Type
Default
Notes

Prompt

Text

Required

Reference Image

Upload

Optional

Used for edits and variations

Size

Select

Square (1:1)

Square (1:1), Landscape (16:9), Landscape (4:3), Landscape (3:2), Portrait (2:3), Portrait (3:4), Portrait (9:16), Ultra-wide (21:9), Tall (9:21)

Resolution

Select

2K

2K, 4K


Modes

Mode
Estimated time
Required inputs
Typical use

Text → Image

~25 seconds

Prompt

High-resolution generation

Image → Image

~25 seconds

Reference Image, Prompt

Inpainting & edits


Output options

Option
Values / notes

Aspect ratio

Square (1:1), Landscape (16:9), Portrait (9:16)

Resolution

2048×2048 (Square), 2048×1152 (Landscape), 1152×2048 (Portrait)


Prompt tips

  • Use photographic tokens for realism.


Last updated

Was this helpful?