Seedream 4.5 Image Edit API

bytedance/seedream/v4.5/edit

1–10 images · JPEG/PNG · 2K/4K · Custom size

Seedream 4.5 Image Edit combines 1–10 reference images with natural language instructions for high-fidelity image editing and multi-image fusion, supporting mask-free refinement, subject identity preservation, and cross-scene synthesis. It executes target modifications accurately while preserving source facial features, core composition, and realistic lighting.

Get API Key

Input

PromptRequired0/3000

Image UrlsRequired0/10

Add images

SizeOptional

NOptional

Enable Safety CheckerOptional

OutputIdle

Generated images will appear here

Estimated cost: —

Continue using

Seedream 4.5 Text to Image Seedream 5.0 Pro Edit Seedance 2.5 Image to Video

Examples

Combine the three reference images into one natural photograph. Use the boy from image 1 with the same face, curly dark hair, freckles, red hooded raincoat with wooden toggles, navy trousers and yellow rubber boots. Use the exact toy sailboat from image 2: teal wooden hull with a white stripe, cream canvas sail with the stitched patch and dark blue painted star, red string at the bow. Place them on the misty loch shoreline from image 3. The boy crouches on the flat rocks at the water's edge, seen in three-quarter side view, gently setting the sailboat onto the calm water with both hands while holding the red string; small ripples spread around the hull. Keep the pebbles, rust bracken hills and misty pines of the location. Soft overcast morning light, consistent scale and shadows, realistic reflections in the water, a quiet tender moment, no text, no watermark.

Edit only the carved text on the two wooden arrow boards. Change the upper board text to exactly "HIDDEN SPRING 2 MI" and the lower board text to exactly "RANGER CABIN 5 MI". Keep the same arrow directions, the same deeply carved cream-painted capital letters, the same font style, letter depth, wood grain, weathering and perspective of each board. Keep everything else unchanged: the post, sagebrush, boulders, red canyon cliffs, sky, lighting and shadows. No other text, no watermark.

Turn this child's crayon drawing into a photograph of a handcrafted miniature diorama built on a tabletop, keeping the exact same composition and layout. The big tree with the round green crown and the wooden treehouse with its rope ladder stays on the left; the family of four sits on the red and white checkered picnic blanket on the right with the basket and watermelon slice; the spotty black-and-white dog runs in the same place; the yellow sun sits in the top right corner with the blue clouds on a painted backdrop; green grass and two red flowers along the bottom. Make everything tangible: sculpted clay figures with simple friendly faces, a real twig tree trunk with felt leaves, a tiny balsa-wood treehouse, fabric blanket, flocked grass. Macro photography with shallow depth of field, soft warm window light, tilt-shift miniature feeling, no text, no watermark.

REST API

Quick Start

This flow submits one valid Seedream 4.5 Edit request, captures its task_id, and reads image URLs after the asynchronous task ends.

Connect to the Vidgo API

Create an API key, keep it on the server, and send it as Authorization: Bearer VIDGO_API_KEY.

Endpoint: POSThttps://api.vidgo.ai/api/generate/submit
Authentication: Authorization: Bearer VIDGO_API_KEY

Submit one image task

Send the current model ID, prompt, and ordered image URLs to the unified submit endpoint. The example keeps only first-call fields.

Submit request

curl --request POST \
  --url "https://api.vidgo.ai/api/generate/submit" \
  --header "Authorization: Bearer $VIDGO_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "bytedance/seedream/v4.5/edit",
  "input": {
    "prompt": "Combine the three reference images into one natural photograph. Use the boy from image 1 with the same face, curly dark hair, freckles, red hooded raincoat with wooden toggles, navy trousers and yellow rubber boots. Use the exact toy sailboat from image 2: teal wooden hull with a white stripe, cream canvas sail with the stitched patch and dark blue painted star, red string at the bow. Place them on the misty loch shoreline from image 3. The boy crouches on the flat rocks at the water's edge, seen in three-quarter side view, gently setting the sailboat onto the calm water with both hands while holding the red string; small ripples spread around the hull. Keep the pebbles, rust bracken hills and misty pines of the location. Soft overcast morning light, consistent scale and shadows, realistic reflections in the water, a quiet tender moment, no text, no watermark.",
    "image_urls": [
      "https://cdn.vidgo.ai/apis/models/bytedance/seedream/v4.5/edit/v1/01/input-01.png",
      "https://cdn.vidgo.ai/apis/models/bytedance/seedream/v4.5/edit/v1/01/input-02.png",
      "https://cdn.vidgo.ai/apis/models/bytedance/seedream/v4.5/edit/v1/01/input-03.png"
    ],
    "size": "3:2",
    "n": 1,
    "enable_safety_checker": true
  }
}'

Wait for the image result

Read data.task_id and poll only while status is not_started or running. Stop on both finished and failed.

Track status

GET https://api.vidgo.ai/api/generate/status/{task_id}

Poll every 2–5 seconds and back off on 429/5xx. Or provide top-level callback_url to receive a terminal-state callback.

not_startedrunningfinishedfailed

Submit response

{
  "code": 200,
  "data": {
    "task_id": "GAPP1TYEWHR5CQVS",
    "status": "running",
    "created_time": "2026-09-25T16:16:53"
  }
}

Finished response

{
  "code": 200,
  "data": {
    "task_id": "GAPP1TYEWHR5CQVS",
    "status": "finished",
    "files": [
      {
        "file_url": "https://cdn.vidgo.ai/apis/models/bytedance/seedream/v4.5/edit/v1/01/output.jpg",
        "file_type": "image"
      }
    ],
    "created_time": "2026-09-25T16:16:53",
    "error_message": null,
    "progress": 100
  }
}

Complete runnable polling example

Expand for the same payload end to end, including business-code, missing-task-ID, failed-state, unexpected-state, and network-error handling.

Complete runnable polling example

set -euo pipefail
: "${VIDGO_API_KEY:?Set VIDGO_API_KEY in your environment}"

REQUEST_BODY=$(cat <<'JSON'
{
  "model": "bytedance/seedream/v4.5/edit",
  "input": {
    "prompt": "Combine the three reference images into one natural photograph. Use the boy from image 1 with the same face, curly dark hair, freckles, red hooded raincoat with wooden toggles, navy trousers and yellow rubber boots. Use the exact toy sailboat from image 2: teal wooden hull with a white stripe, cream canvas sail with the stitched patch and dark blue painted star, red string at the bow. Place them on the misty loch shoreline from image 3. The boy crouches on the flat rocks at the water's edge, seen in three-quarter side view, gently setting the sailboat onto the calm water with both hands while holding the red string; small ripples spread around the hull. Keep the pebbles, rust bracken hills and misty pines of the location. Soft overcast morning light, consistent scale and shadows, realistic reflections in the water, a quiet tender moment, no text, no watermark.",
    "image_urls": [
      "https://cdn.vidgo.ai/apis/models/bytedance/seedream/v4.5/edit/v1/01/input-01.png",
      "https://cdn.vidgo.ai/apis/models/bytedance/seedream/v4.5/edit/v1/01/input-02.png",
      "https://cdn.vidgo.ai/apis/models/bytedance/seedream/v4.5/edit/v1/01/input-03.png"
    ],
    "size": "3:2",
    "n": 1,
    "enable_safety_checker": true
  }
}
JSON
)

SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --request POST \
  --url "https://api.vidgo.ai/api/generate/submit" \
  --header "Authorization: Bearer $VIDGO_API_KEY" \
  --header "Content-Type: application/json" \
  --data "$REQUEST_BODY")

TASK_ID=$(printf '%s' "$SUBMIT_RESPONSE" | jq -r '.data.task_id // .task_id // empty')
if [ -z "$TASK_ID" ]; then
  printf 'Submit response did not include task_id:
%s
' "$SUBMIT_RESPONSE" >&2
  exit 1
fi

while true; do
  STATUS_RESPONSE=$(curl --silent --show-error --fail-with-body \
    --url "https://api.vidgo.ai/api/generate/status/$TASK_ID" \
    --header "Authorization: Bearer $VIDGO_API_KEY")
  STATUS=$(printf '%s' "$STATUS_RESPONSE" | jq -r '.data.status // .status // empty')

  case "$STATUS" in
    finished)
      printf '%s' "$STATUS_RESPONSE" | jq -r '(.data.files // .files // [])[]?.file_url'
      break
      ;;
    failed)
      printf '%s' "$STATUS_RESPONSE" | jq -r '.data.error_message // .error_message // "Generation failed"' >&2
      exit 1
      ;;
    not_started|running)
      sleep 2
      ;;
    *)
      printf 'Unexpected task status: %s
' "$STATUS" >&2
      exit 1
      ;;
  esac
done

Request Parameters

This table lists only public fields inside input. See Quick Start for top-level model and optional callback_url usage.

Field	Type	Required	Default	Description
input.prompt	string	Yes	–	Describe the desired modifications in 1–3,000 characters, referencing input images as "image 1", "image 2", etc.
input.image_urls	string[]	Yes	–	1–10 ordered HTTP(S) reference image URLs. Formats: JPEG, PNG.
input.size	string \| object	No	1:1	Output size: 2K, 4K, or an aspect ratio. For a custom size, enter WIDTHxHEIGHT (for example 1920x4096) or {"width":2304,"height":3072} with positive integers. Defaults to 1:1.
input.n	integer	No	1	Number of images to generate: integer 1–15. Defaults to 1.
input.enable_safety_checker	boolean	No	true	Set true to enable safety checking or false to turn it off. Defaults to true.

Response Fields

The submit response identifies the task. The status response supplies image files or failure details at a terminal state.

Field	Type	Description
code	integer	Business response code; 0 or 200 indicates success.
message	string	Optional human-readable response message.
data.task_id	string	Unique task ID used for status queries after submission.
data.status	string	not_started, running, finished, or failed.
data.created_time	string	Task creation timestamp.
data.progress	integer	Progress from 0–100 in status responses.
data.files[]	array	Generated files in returned order when finished.
data.files[].file_url	string	Direct generated-image URL.
data.files[].file_type	string	Image results are identified as image.
data.files[].watermark_url	string \| null	Optional watermarked-file URL when present.
data.error_message	string \| null	Readable failure reason for failed tasks.

Task Lifecycle

Only the first two states may be polled again. finished and failed are terminal.

not_started
Accepted and queued; wait before checking again.
running
Image generation is in progress; continue polling.
finished
Successful terminal state; validate files and read image URLs.
failed
Failure terminal state; read error_message and stop.

Polling and Errors

AuthenticationKeep a valid VIDGO_API_KEY on the server and send requests with Bearer authentication.
Validation400 indicates an invalid model ID, field, enum, length, or image count.
Polling intervalStart at 2–5 seconds, use exponential backoff for 429/5xx, and enforce an overall timeout.
Terminal statesRead image URLs from files after finished, or error_message after failed. Both states end polling.
Callback optionProvide a public top-level callback_url to receive a flat task object after finished or failed.
Network errorsDistinguish request timeout, network failure, and task failure. Retry the same query only when safe.

Model Specifications

Specification	Value	Details
Model	bytedance/seedream/v4.5/edit	Edit reference images using text instructions.
image_urls	1–10	JPEG / PNG
size	1:1 / 4:3 / 3:4 / 16:9 / 9:16 / 3:2 / 2:3 / 21:9 / 2K / 4K	Default: 1:1. Custom WIDTHxHEIGHT or {"width","height"} is also accepted.
n	1–15	Default: 1.

Seedream 4.5 Edit API documentation

Seedream 4.5 Image Edit

Seedream 4.5 Image Edit is a flagship multimodal image editing and reference-guided fusion model developed by the ByteDance Seed team. Sharing a unified architecture with its companion text-to-image endpoint and eliminating the need for manual mask painting (Mask-free), it enables precise local modifications, micro-material transitions (such as shifting metallic textures into transparent liquid), and cross-image synthesis via natural language instructions. The model natively accepts 1 to 10 ordered public reference image URLs, allowing users to assign clear roles—such as subject, apparel, props, or background—to "image 1", "image 2", and so on directly within prompts. Powered by advanced subject segmentation and attribute disentanglement, Seedream 4.5 executes complex compositional edits while faithfully preserving facial identity, perspective geometry, and authentic ambient lighting for e-commerce, commercial design, and digital creative workflows.

Why Choose Seedream 4.5 Image Edit API?

1–10 Reference Multi-Image Fusion and Role AssignmentIngests up to 10 ordered reference images with explicit role assignment—such as subject, apparel, props, or background—seamlessly synthesizing multi-asset campaigns with zero material fees.
Mask-Free Natural Language Editing and Material TransformationEliminates manual brushwork and lasso cutouts, executing precise local edits, background replacements, and advanced material attribute changes (such as shifting metallic textures into transparent liquid) via text prompts alone.
High-Fidelity Subject Identity and Ambient Lighting RetentionDisentangles identity attributes from environmental elements, reliably locking facial contours, micro-expressions, and natural light reflections during extensive background or outfit edits.
1–15 Batch Edit Variations with Adaptive Aspect RatiosOutputs up to 15 diverse edit and composition variations per task across 2K/4K and 8 preset aspect ratios, intelligently outpainting or reframing backgrounds to fit multi-platform layouts.

Parameters

Parameter	Requirement	Description
prompt	Required	Describe the desired modifications in 1–3,000 characters, referencing input images as "image 1", "image 2", etc.
image_urls	Required	1–10 ordered HTTP(S) reference image URLs. Formats: JPEG, PNG.
size	Optional	Output size: 2K, 4K, or an aspect ratio. For a custom size, enter WIDTHxHEIGHT (for example 1920x4096) or {"width":2304,"height":3072} with positive integers. Defaults to 1:1. Default`1:12K4K4:33:416:99:163:22:321:9`
n	Optional	Number of images to generate: integer 1–15. Defaults to 1. Default`123456789101112131415`
enable_safety_checker	Optional	Set true to enable safety checking or false to turn it off. Defaults to true. Default`true`

How to Use

Prepare and Provide Reference ImagesSupply 1 to 10 publicly accessible HTTP(S) image URLs pointing to clear JPEG or PNG assets (or upload directly in the interactive playground).
Write Explicit Editing InstructionsReference source inputs using "image 1" and "image 2", specifying what to keep and what to alter (for example, "keep the person from image 1 and place them on the beach from image 2").
Submit Task and Retrieve Blended AssetsSpecify target aspect ratios and output image count n, then submit your asynchronous request to track progress and receive the high-fidelity output URLs.

Pricing

Each generation costs 5 credits ($0.025). n is the number of generated images; the total is 5 × n credits.

Usage	Rate	Details
Standard generation	5 credits / generation · $0.025 / generation	5 × n credits · $0.025 × n

Best Use Cases

E-Commerce Background and Scene ReplacementTransplant studio product and model photography into outdoor landscapes or seasonal settings with realistic light interaction and shadow cast.
Multi-Asset Character and Product SynthesisExtract a character face from image 1, apparel from image 2, and background from image 3 to produce unified, coherent visual campaigns.
Mask-Free Portrait Retouching and StylingUpdate character hairstyles, clothing details, or hand-held props with simple text instructions while keeping facial identities fully intact.

Pro Tips

Explicit role assignment: Assign clear responsibilities in your prompt (such as "image 1 for subject, image 2 for background, image 3 for apparel") to guide multi-asset feature extraction.
Adopt "preserve X while modifying Y" syntax: Clearly designate unchanged elements (such as "preserve the facial identity in image 1") to keep model focus balanced.
Align reference lighting and clarity: When combining elements from multiple images, choosing inputs with similar lighting direction and clarity yields more natural composite results.

Notes

Reference images required: This endpoint requires 1 to 10 public reference image URLs. For pure text-to-image synthesis, use Seedream 4.5 Text to Image.
Public URL accessibility: Reference images must be accessible over HTTP(S) without basic authentication or private network restrictions, in JPEG or PNG format.
No per-image surcharge: Providing up to 10 references incurs no extra material fees; billing remains flat at 5 credits ($0.025) per successfully generated image.

Related Models

Grok Imagine Image 2.0 Text to Image preview

Grok Imagine Image 2.0 Text to ImageGenerate crisp, high-fidelity images up to 2K from text prompts with Grok Imagine Image 2.0. Features designer-grade typography, precise prompt adherence, and versatile aspect ratios.

Grok Imagine Image 2.0 EditGenerate crisp, high-fidelity images up to 2K from text prompts with Grok Imagine Image 2.0. Features designer-grade typography, precise prompt adherence, and versatile aspect ratios.

Qwen Image 3.0 Text-to-ImageAlibaba Qwen Image 3.0 Standard and Pro generation and editing with eight aspect ratios, 1K or 2K output, and ordered one-to-three-image editing.

Qwen Image 3.0 Image EditAlibaba Qwen Image 3.0 Standard and Pro generation and editing with eight aspect ratios, 1K or 2K output, and ordered one-to-three-image editing.

Qwen Image 3.0 Pro Text-to-Image preview

Qwen Image 3.0 Pro Text-to-ImageAlibaba Qwen Image 3.0 Standard and Pro generation and editing with eight aspect ratios, 1K or 2K output, and ordered one-to-three-image editing.

Qwen Image 3.0 Pro Image EditAlibaba Qwen Image 3.0 Standard and Pro generation and editing with eight aspect ratios, 1K or 2K output, and ordered one-to-three-image editing.

Seedream 5.0 Pro Text-to-ImageThe ByteDance Seedream 5.0 Pro API generates 1K or 2K JPEG/PNG images from prompts and edits one to ten ordered HTTP(S) image inputs.

Seedream 5.0 Pro Image EditThe ByteDance Seedream 5.0 Pro API generates 1K or 2K JPEG/PNG images from prompts and edits one to ten ordered HTTP(S) image inputs.

Nano Banana 2 Lite Text to Image preview

Nano Banana 2 Lite Text to ImageGenerate crisp 1K images in ~4 seconds from text prompts with Nano Banana 2 Lite. Optimized for high-volume rapid iteration, in-image text in 25+ languages, and 10 aspect ratios at $0.025 per generation.

Nano Banana 2 Lite EditGenerate crisp 1K images in ~4 seconds from text prompts with Nano Banana 2 Lite. Optimized for high-volume rapid iteration, in-image text in 25+ languages, and 10 aspect ratios at $0.025 per generation.

Grok Imagine Image Quality Text to Image preview

Grok Imagine Image Quality Text to ImageGenerate high-fidelity images up to 2K resolution from text prompts with Grok Imagine Image Quality. Features photorealistic detail, sharp text rendering, and 5 aspect ratios.

Grok Imagine Image Quality EditGenerate high-fidelity images up to 2K resolution from text prompts with Grok Imagine Image Quality. Features photorealistic detail, sharp text rendering, and 5 aspect ratios.

GPT Image 2 Text-to-ImageOpenAI image generation and editing with quality tiers, flexible size controls, and single-image outputs up to 4K.

GPT Image 2 Image EditOpenAI image generation and editing with quality tiers, flexible size controls, and single-image outputs up to 4K.

Wan 2.7 Image Standard Text to Image preview

Wan 2.7 Image Standard Text to ImageCreate and edit images from prompts or reference photos with Wan 2.7 Image Standard and Pro models.

Wan 2.7 Image Standard EditCreate and edit images from prompts or reference photos with Wan 2.7 Image Standard and Pro models.

Seedream 4.5 Image Edit API frequently asked questions

What is the Seedream 4.5 Image Edit API?

Seedream 4.5 Image Edit is a ByteDance Seed model for high-quality image editing and multi-image reference fusion. It combines 1–10 reference images with natural language instructions to perform precise mask-free local editing, multi-image feature synthesis, and subject identity preservation. Built on an advanced global scaling visual understanding and unified generation architecture, it executes requested changes while faithfully preserving source facial features, composition, and authentic lighting. You can call it programmatically or try it from the playground above.

How many reference images does Seedream 4.5 Image Edit support?

It supports 1 to 10 public reference images per request. Images are passed as an ordered array and cited in the prompt as "image 1" through "image 10", with no additional cost for uploading multiple references.

Is manual masking required to use Seedream 4.5 Image Edit?

No. The model provides native mask-free editing through natural language understanding, identifying target regions and blending modifications seamlessly without requiring manual brush masks or alpha cutouts.

How does Seedream 4.5 Image Edit maintain subject identity consistency?

The model separates identity features from clothing and environment in latent space. Explicit instructions like "preserve the face and hair of image 1" further stabilize character consistency across different composite scenes.

Does providing multiple reference images increase the cost for Seedream 4.5 Image Edit?

No. Billing is determined strictly by the number of output images generated at 5 credits ($0.025) each, whether you supply 1 reference image or the maximum of 10 references.

Can Seedream 4.5 Image Edit change the image aspect ratio or resolution?

Yes. You can select 2K, 4K, or any of the eight preset aspect ratios, or pass custom dimensions in the size parameter. The model intelligently extends or reframes the background to fit the new canvas dimensions.

Can Seedream 4.5 Image Edit generate multiple edit variations at once?

Yes. By configuring the n parameter from 1 to 15, you can generate up to 15 diverse editing and composition variations in a single run, billed only for the images successfully produced.