Bytedance Seedream V4.5 Sequential

Bytedance Seedream V4.5 Sequential

Playground

Seedream 4.5 Sequential generates multi-image sets with consistent characters and objects, unifying palette, lighting, and style across all outputs. Supports up to 4K results for campaigns, storyboards, and product lines. Ready-to-use REST inference API, best performance, no cold starts, affordable pricing.

Features

Seedream 4.5 Sequential is ’s multi-image generation model for creating whole series of images in one go. It keeps characters, props, and style consistent across all outputs, making it ideal for KVs, comic panels, and any visual set that should “look like one universe”.

Model highlights

• Character consistency – Locks onto the same character identity (face, hairstyle, body shape) across all generated frames.
• Object & prop stability – Reuses key objects (products, logos, props) so they don’t randomly change between images.
• Unified visual style – Maintains palette, lighting, camera feel, and rendering style across the whole set.
• Multi-image output – Generate several images in a single request, all driven by the same prompt.
• 4K-ready detail – Supports resolutions up to 8192 × 8192 per image for hero KVs and print-adjacent work.
• Typography aware – Strong on-image text rendering for branded content, titles, and UI-like elements.

Best suited for

• Comic strips and story panels with recurring characters
• Brand KV series and campaign sets built around the same hero figure
• Product lineup / colourway visualisation
• Storyboard or animatic keyframes
• Social media content series that must feel coherent in the grid
• Multi-step marketing journeys (awareness → consideration → conversion visuals)

Pricing

Billing is per generated image, controlled by max_images.

• $0.04 per image
• Formula: total_price = $0.04 × max_images

Example costs:

The exact price for your chosen settings is always shown in the WaveSpeedAI interface before you run the job.

How to use

1. Enter your prompt Describe the scene, characters, and what must stay consistent, e.g. “Same girl with red hoodie and headphones, different city locations, cinematic lighting.”

2. Set max_images Choose how many images you want in the series. Each one will follow the same prompt and consistency logic.

3. Set size Choose width and height, up to 8192 × 8192 for maximum detail.

4. Run and review Generate the series, check continuity across faces, outfits, and props, then refine the prompt for another pass if needed.

Notes

• Please set the max_image first, and then input how many images you want to generate in prompt! Such as:

• max_image = 4.

Prompt: I want to generate 4 images… + (your prompt)

Model comparison & related tools

• Nano Banana Pro Google’s ultra-low-cost, high-throughput T2I model. Great for lots of individual images or quick ideation, but it doesn’t provide built-in cross-image character locking.

• Seedream V4 – sequential ’s single-image Seedream generator with rich detail and stylish output. Ideal when you want one-off hero shots or standalone illustrations rather than a strictly consistent series.

• Qwen Image Edit Plus Qwen-Image is a 20B MMDiT-based text-to-image generation model, especially strong at native text rendering in both English and Chinese. It is a powerful creative tool for posters, comics, and visual storytelling, while also excelling at general image generation from photorealism to anime.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "size": "2048*2048",
  "max_images": 1
}
JSON
)

# 1. Submit the prediction.
SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  -X POST "https://api.wavespeed.ai/api/v3/bytedance/seedream-v4.5/sequential" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}")

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id
' >&2
  exit 1
fi
RESULT_URL=$(printf '%s' "${TASK}" | jq -r '.urls.get // empty')
if [ -z "${RESULT_URL}" ]; then RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"; fi

# 2. Poll until the prediction finishes.
while true; do
  RESPONSE=$(curl --silent --show-error --fail-with-body \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}")
  RESULT=$(printf '%s' "${RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout) printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    created|processing) sleep 2 ;;
    *) printf 'Unexpected status: %s
' "${STATUS}" >&2; exit 1 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Response Parameters

Result Request Parameters

Result Response Parameters