概述
字节 Seedream 系列图像生成模型，支持文生图、图生图、多图融合、批量生成和流式输出。支持的模型
| 模型 | 特点 |
|---|---|
seedream-4.5 | 最新版本，功能最全，支持高分辨率、图生图、多图融合、批量生成、流式输出 |
seedream-4.0 | 稳定版本，支持高分辨率、图生图、多图融合 |
seedream-3.0-t2i | 基础文生图，仅支持文本生成图像 |
官方API文档
查看字节官方API参数说明
请求参数
基础参数
| 参数 | 类型 | 必填 | 说明 |
|---|---|---|---|
model | string | 是 | 模型名称：seedream-3.0-t2i、seedream-4.0、seedream-4.5 |
prompt | string | 是 | 图像生成提示词，最多 2000 tokens |
size | string | 否 | 输出图像尺寸，支持简化格式或像素格式（见下方说明） |
n | integer | 否 | 生成图片数量，范围 1-15，默认 1 |
watermark | boolean | 否 | 是否添加水印，默认 true |
prompt_priority | string | 否 | 提示词优化策略：standard（标准模式，质量更高，处理时间更长） |
尺寸参数 (size)
支持两种格式：| 格式 | 说明 | 示例 |
|---|---|---|
| 简化格式 | 模型自动确定最佳尺寸 | 2K、4K |
| 像素格式 | 显式指定宽×高 | 2048x2048、2560x1440、4096x4096 |
使用简化格式时，可在 prompt 中描述宽高比/形状/用途，模型会自动选择最佳分辨率
支持的分辨率
以下是 Seedream 4.0/4.5 常用分辨率对照表：| 宽高比 | 像素尺寸 (宽×高) |
|---|---|
| 1:1 | 2048×2048 |
| 4:3 | 2304×1728 |
| 3:4 | 1728×2304 |
| 16:9 | 2560×1440 |
| 9:16 | 1440×2560 |
图片限制
| 限制项 | 说明 |
|---|---|
| 宽高比范围 | [1/16, 16] |
| 最小宽高 | 大于 14 像素 |
| 最大总像素 | 不超过 6000×6000 |
| 单张图片大小 | 最大 10MB |
| 支持格式 | .jpeg, .jpg, .png, .webp, .bmp, .tiff, .gif |
| 最大参考图数量 | 14 张 |
| 参考图+生成图总数 | ≤ 15 张 |
图生图参数
| 参数 | 类型 | 必填 | 说明 |
|---|---|---|---|
image / image_urls | string/array | 否 | 参考图像URL，单图为字符串，多图为数组（最多14张） |
sequential_image_generation | string | 否 | 图像生成模式：disabled（单图输出）、auto（批量生成） |
sequential_image_generation_options | object | 否 | 批量生成选项，需配合 sequential_image_generation: "auto" |
sequential_image_generation_options.max_images | integer | 否 | 批量生成的最大图片数量 |
流式输出参数
| 参数 | 类型 | 必填 | 说明 |
|---|---|---|---|
stream | boolean | 否 | 是否启用流式输出，默认 false |
回调参数
| 参数 | 类型 | 必填 | 说明 |
|---|---|---|---|
callback_url | string | 否 | 任务完成后的 HTTPS 回调地址（仅 HTTPS，最长 2048 字符，超时 10s，最多重试 3 次） |
文生图用例
seedream-3.0-t2i
基础文生图，无需复杂配置：curl https://api.aigc.it/v1/images/generations \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <YOUR_TOKEN>" \
-d '{
"model": "seedream-3.0-t2i",
"prompt": "a cat"
}'
seedream-4.0 / seedream-4.5
支持高分辨率和水印控制：curl https://api.aigc.it/v1/images/generations \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <YOUR_TOKEN>" \
-d '{
"model": "seedream-4.5",
"prompt": "a cat",
"size": "2K",
"watermark": false
}'
{
"model": "seedream-4-5-251128",
"created": 1757323224,
"data": [
{
"url": "https://...",
"size": "1760x2368"
}
],
"usage": {
"generated_images": 1,
"output_tokens": 16280,
"total_tokens": 16280
}
}
图生图用例
基于单张参考图进行重绘或变体生成。 核心参数：image：单张图片URL（字符串）sequential_image_generation：设为"disabled"表示单图输出
curl https://api.aigc.it/v1/images/generations \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <YOUR_TOKEN>" \
-d '{
"model": "seedream-4.5",
"prompt": "Keep the model pose... Change clothing to transparent water...",
"image": "https://ark-doc.tos-ap-southeast-1.bytepluses.com/doc_image/seedream4_5_imageToimage.png",
"size": "2K",
"watermark": false
}'
{
"model": "seedream-4-5-251128",
"created": 1757323224,
"data": [
{
"url": "https://...",
"size": "1760x2368"
}
],
"usage": {
"generated_images": 1,
"output_tokens": 16280,
"total_tokens": 16280
}
}
多图融合用例
多张参考图生成单张结果（如：人物+服装=换装效果）。 核心参数：image：多张图片URL（数组）sequential_image_generation：设为"disabled"表示融合输出单图
curl https://api.aigc.it/v1/images/generations \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <YOUR_TOKEN>" \
-d '{
"model": "seedream-4.5",
"prompt": "Replace the clothing in image 1 with the outfit from image 2.",
"image": [
"https://ark-doc.tos-ap-southeast-1.bytepluses.com/doc_image/seedream4_5_imageToimage.png",
"https://ark-doc.tos-ap-southeast-1.bytepluses.com/doc_image/seedream4_5_clothing.png"
],
"sequential_image_generation": "disabled",
"size": "2K",
"watermark": false
}'
{
"model": "seedream-4-5-251128",
"created": 1757323224,
"data": [
{
"url": "https://...",
"size": "1760x2368"
}
],
"usage": {
"generated_images": 1,
"output_tokens": 16280,
"total_tokens": 16280
}
}
批量生成用例
一次生成多张图片，适合需要多个变体的场景。 核心参数：sequential_image_generation：设为"auto"启用批量生成sequential_image_generation_options.max_images：指定生成数量
curl https://api.aigc.it/v1/images/generations \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <YOUR_TOKEN>" \
-d '{
"model": "seedream-4.5",
"prompt": "Generate 3 images of a girl riding a roller coaster...",
"image": [
"https://ark-doc.tos-ap-southeast-1.bytepluses.com/doc_image/seedream4_5_imageToimage.png"
],
"sequential_image_generation": "auto",
"sequential_image_generation_options": {
"max_images": 3
},
"size": "2K",
"watermark": false
}'
{
"model": "seedream-4-5-251128",
"created": 1757388756,
"data": [
{ "url": "https://...", "size": "2720x1536" },
{ "url": "https://...", "size": "2720x1536" },
{ "url": "https://...", "size": "2720x1536" }
],
"usage": {
"generated_images": 3,
"output_tokens": 48960,
"total_tokens": 48960
}
}
流式输出用例
使用 Server-Sent Events (SSE) 实时返回生成进度和结果。 核心参数：stream：设为true启用流式输出- 配合
sequential_image_generation: "auto"使用效果最佳
curl https://api.aigc.it/v1/images/generations \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <YOUR_TOKEN>" \
-d '{
"model": "seedream-4.5",
"prompt": "Generate four images of different poses...",
"image": "https://ark-doc.tos-ap-southeast-1.bytepluses.com/doc_image/seedream4_5_imageToimage.png",
"sequential_image_generation": "auto",
"sequential_image_generation_options": {
"max_images": 4
},
"size": "2K",
"stream": true,
"watermark": false
}'
event: image_generation.partial_succeeded
data: {
"type": "image_generation.partial_succeeded",
"model": "seedream-4-5-251128",
"created": 1757396757,
"image_index": 0,
"url": "https://...",
"size": "2496x1664"
}
event: image_generation.partial_succeeded
data: {
"type": "image_generation.partial_succeeded",
"model": "seedream-4-5-251128",
"created": 1757396785,
"image_index": 1,
"url": "https://...",
"size": "2496x1664"
}
event: image_generation.completed
data: {
"type": "image_generation.completed",
"model": "seedream-4-5-251128",
"created": 1757396825,
"usage": {
"generated_images": 2,
"output_tokens": 48672,
"total_tokens": 48672
}
}
data: [DONE]