ByteDance Seedream 4.5 image editing: unified generation-editing architecture with up to 14 reference images for edits and composites, consistent subjects, 2K/4K output.
Provider: ByteDance
Category: image generation
Endpoint: /v1/tasks
Status: Available
Cost: See live page pricing
ByteDance Seedream 4.5 image editing: unified generation-editing architecture with up to 14 reference images for edits and composites, consistent subjects, 2K/4K output.
Pricing
Standard Usage
Ready for generation
Configure your parameters and click "Run" to see the output here.
Real API editing results
Explore product recoloring, multi-reference composites, relighting, character scene changes, and lifestyle edits. Drag the divider on single-image edits and copy the exact prompt used for each result.
Changes the material colors while preserving the shoe shape, stitching, camera angle, shadow, and background.
recolor the sneaker to a deep forest green canvas with an off-white sole and matching green laces; keep the exact same shoe shape, stitching, camera angle, lighting, shadow and gray background unchanged
Image 1 · Product
Image 2 · Material
Image 3 · Scene
ResultUses three ordered references to rebuild the sneaker in cable-knit wool and place it on a foggy forest trail.
rebuild the sneaker from the first image with its canvas upper made of the burnt-orange cable-knit wool from the second image, then place it as a product hero shot on the wet leaf-covered forest trail from the third image, low camera angle, soft morning fog, keep the shoe's shape and proportions from the first image, photorealistic
Rebuilds the directional light while preserving the bowl, toppings, camera angle, and composition.
relight this exact bowl of ramen as a bright editorial food photograph: soft window key light from the upper left, clean neutral white balance, gentle steam rising, richer broth and egg color, keep the same bowl, ingredients, table and overhead framing
Moves the fisherman from the harbor into a boat wheelhouse while retaining his face, beard, beanie, and sweater.
move this same fisherman, with identical face, beard, beanie and green sweater, into the wheelhouse of a small fishing boat at dawn, standing at the wooden helm, cold blue morning light through the windows mixing with warm cabin light, photorealistic
Moves the ceramic coffee dripper into a warm kitchen story with coordinated props and golden-hour light.
E-commerce lifestyle edit: move this matte ceramic pour-over coffee dripper onto a warm reclaimed-wood kitchen counter with a steaming mug and a burlap coffee bean bag, golden-hour lighting, keep the product shape and matte ceramic finish recognizable, no text
Choose an adjacent model by input type, editing preference, and workflow.
Seedream 4.5 is ByteDance's unified generation-and-editing image model. This endpoint accepts 1–14 reference images for material swaps, scene changes, seasonal transformations, multi-image composites, and commercial edits that preserve subjects and in-image text.
The Playground and API use the same model ID: seedream-4.5/image-to-image. Upload reference images to validate an edit instruction, then move the approved parameters into an asynchronous task workflow.
These specifications reflect the current HiAPI Seedream 4.5 Image-to-Image request contract.
The current 2K and 4K tiers share one public rate shown by the live page price; the backend exposes no separate resolution surcharge.
This workflow follows the model capability: image editing emphasizes reference order, change-and-preserve constraints, and result review instead of a generic text-to-image sequence.
This is the minimum request for an asynchronous image-editing task. Use the API docs for callbacks, task retrieval, and full parameter definitions.
POST /v1/tasksseedream-4.5/image-to-imagecurl --request POST 'https://api.hiapi.ai/v1/tasks' \
--header 'Authorization: Bearer YOUR_API_KEY' \
--header 'Content-Type: application/json' \
--data '{
"model": "seedream-4.5/image-to-image",
"input": {
"prompt": "recolor the sneaker to a deep forest green canvas with an off-white sole and matching green laces; keep the exact same shoe shape, stitching, camera angle, lighting, shadow and gray background unchanged",
"image_urls": [
"https://static.hiapi.ai/blog/seedream-4-5-image-to-image-prompt-recipes/src-sneaker.jpg"
],
"aspect_ratio": "3:2",
"resolution": "2K"
}
}'The create request immediately returns a taskId; image editing continues asynchronously.
The image-to-image endpoint uses uploaded images as editing references for recoloring, scene changes, relighting, style transfer, and multi-image composition. Text-to-image starts from a written prompt without a source image. Use this image-to-image model when subjects, composition, or in-image text should remain recognizable.
Each request accepts 1–14 JPEG, PNG, or WebP images up to 10 MB each; SVG is not supported. For multi-image jobs, refer to inputs as “image 1” and “image 2” in the prompt, matching their order in the image_urls array.
Billing is per successful output image, not per reference image. The current public rate is the same for 2K and 4K output. See the live model price and task record for the final charge.
State both what should change and what must remain unchanged, such as the subject silhouette, camera angle, contact shadow, storefront layout, or calligraphy. Review the result in the Playground first. Keep human review for brand text or layouts that must be exact.
Seedream 4.5 accepts up to 14 references and is a strong fit for multi-image composition, Chinese scenes, and edits that preserve in-image text. FLUX.2 Image-to-Image accepts up to 8 references and provides an alternative for photoreal consistency and editing style. Test both with the same assets when the choice matters.
Send a POST request to /v1/tasks, set model to seedream-4.5/image-to-image, and pass reference images in input.image_urls. The create request returns a taskId while generation continues asynchronously. See the API docs for task retrieval, callbacks, errors, and the full parameter contract.
Add 1–14 JPEG, PNG, or WebP images. For multi-image jobs, keep image 1 and image 2 aligned with upload order.
Describe the edit, then list the subject, text, angle, lighting, or composition that must remain unchanged.
Inspect details in the Playground, then create async tasks with the same model ID and retrieve results by taskId.