bytedance/seedream-5-lite

Seedream 5.0 lite: image generation with built-in reasoning, example-based editing, and deep domain knowledge

4.1K runs
Playground API Examples README

Seedream 5.0 lite

Seedream 5.0 lite is ByteDance’s latest image generation model. It goes beyond standard text-to-image by adding multi-step reasoning, example-based editing, and deep domain knowledge to the generation process.

What’s new in 5.0

Example-based editing

Instead of describing a complex edit in words, show the model what you want. Give it a before/after pair, then a new image — the model figures out what changed and applies the same transformation. This works for material swaps, style transfers, scene changes, and more.

Logical reasoning

Seedream 5.0 reasons through spatial relationships, physics, and processes. Ask it to put objects on a seesaw with correct weight distribution, draw a clock with hands in the right positions, or illustrate a metamorphosis across life stages — it gets the details right.

Deep domain knowledge

The model understands professional conventions across architecture, science, health, and design. Feed it a floor plan sketch and get a photorealistic interior rendering. Ask for a scientific cross-section diagram and get labeled, accurate illustrations.

Features

  • Text-to-image: Generate images from text prompts with strong aesthetic quality
  • Image-to-image: Edit existing images using natural language instructions
  • Multi-image blending: Combine up to 14 reference images with text to create new compositions
  • Sequential batch generation: Generate sets of related images (storyboards, brand identity packages, character sheets) in one request
  • Text rendering: Accurate typography with support for multiple languages — wrap text in double quotes for best results
  • Output format: PNG or JPEG output

Resolutions

  • 2K: Up to 2048px base dimension
  • 3K: Up to 3072px base dimension

Supported aspect ratios: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 21:9

Prompting tips

  1. Use natural language, not keyword lists. “A girl in a lavish dress walking under a parasol along a tree-lined path, in the style of a Monet oil painting” works better than “girl, umbrella, tree-lined street, oil painting texture.”
  2. Use double quotes for text rendering. If you want specific text in your image, wrap it in double quotation marks.
  3. Be specific about what to keep. When editing, tell the model what shouldn’t change: “Replace the hat with a crown, keeping the pose and expression unchanged.”
  4. For example-based editing, show don’t tell. When the transformation is hard to describe in words, provide a before/after example pair as input images.
  5. Specify your use case. Telling the model “Design a logo for a gaming company” gives better results than just describing the visual elements.

API quickstart

import replicate

output = replicate.run(
    "bytedance/seedream-5",
    input={
        "prompt": "A color film-inspired portrait with shallow depth of field, fine grain suggesting high ISO film stock, candid documentary style",
        "size": "2K",
        "aspect_ratio": "3:2",
    }
)

print(output)

With image input

output = replicate.run(
    "bytedance/seedream-5",
    input={
        "prompt": "Transform the color grading to match a Wong Kar-wai film — saturated teal shadows, warm amber highlights, soft diffusion",
        "image_input": ["https://example.com/portrait.jpg"],
        "size": "2K",
    }
)

Batch generation

output = replicate.run(
    "bytedance/seedream-5",
    input={
        "prompt": "A series of 4 coherent illustrations of a courtyard across the four seasons",
        "sequential_image_generation": "auto",
        "max_images": 4,
    }
)
Model created
Model updated