Generate images (compatible with OpenAI Dall-E)

OpenAI Dall-E-style image generation API.

• POST /v1/images/generations: Create an image generation task

Standard authentication headers:

• Content-Type: application/json

• model: string: Required, public model name

• prompt: string: Recommended, image generation prompt

• n: number: Optional, number of images to generate

• size: string: Optional, size or resolution specification

• quality: string: Optional, quality level

• response_format: string: Optional, response format

Common compatible extension parameters:

• image
• image_url
• image_urls
• image_list
• subject_image_list
• scene_image
• style_image
• negative_prompt
• aspect_ratio
• resolution
• callback_url
• external_task_id

Kling Dedicated extension parameters:

• model_name
• element_list
• result_type
• series_amount
• human_image
• cloth_image
• up_expansion_ratio
• down_expansion_ratio
• left_expansion_ratio
• right_expansion_ratio

3.1 Nano Banana Series

Public model:

• nano-banana
• nano-banana-pro
• nano-banana-pro-2k
• nano-banana-pro-4k
• nano-banana-2
• nano-banana-2-2k
• nano-banana-2-4k

Capabilities:

• nano-banana: Basic text-to-image, also compatible with image editing
• nano-banana-pro*: Higher-quality version
• nano-banana-2*: Newer model family
• *-2k / *-4k: High-resolution variants

Common parameters:

• Text-to-image: model + prompt
• Image editing: model + prompt + image, also compatible with image_urls
• Common extensions: n size quality response_format

Example:

{
  "model": "nano-banana",
  "prompt": "a cute cat running in cloud",
  "n": 1,
  "size": "1024x1792"
}

{
  "model": "nano-banana-pro",
  "prompt": "把主体改成赛博朋克夜景",
  "image": "https://example.com/input.png",
  "n": 1
}

3.2 Seedream Series

Public model:

• doubao-seedream-3-0-t2i-250415
• doubao-seedream-4-0-250828
• doubao-seedream-4-5-251128
• doubao-seedream-5-0-260128

Capabilities:

• Use the text-to-image call uniformly
• It is recommended to pass the standard parameter first: prompt + model + size/aspect_ratio

Common parameters:

• model + prompt
• Optional: size aspect_ratio n

3.3 Kling Standard Image Main Model

Public base model:

• kling-image

Required:

• model
• model_name

Supported model_name:

• kling-v1
• kling-v1-5
• kling-v2
• kling-v2-new
• kling-v2-1
• kling-v3

Capabilities:

• Text-to-image: model + model_name + prompt
• Image-to-image: add image on top of text-to-image
• Multi-image generation: pass subject_image_list, or pass multiple image_urls/images
• Outpainting: pass image and the outpainting ratio parameter

Common parameters:

• image
• subject_image_list
• scene_image
• style_image
• aspect_ratio
• resolution
• up_expansion_ratio
• down_expansion_ratio
• left_expansion_ratio
• right_expansion_ratio

Example:

{
  "model": "kling-image",
  "model_name": "kling-v3",
  "prompt": "海边日落，电影感人像",
  "aspect_ratio": "16:9",
  "n": 1
}

3.4 Kling Omni Image Main Model

Public base model:

• kling-omni-image

Required:

• model
• model_name

Supported model_name:

• kling-image-o1
• kling-v3-omni

Common parameters:

• image_list
• element_list
• resolution
• aspect_ratio
• result_type
• series_amount

Example:

{
  "model": "kling-omni-image",
  "model_name": "kling-v3-omni",
  "prompt": "生成一张海边跳舞的图片",
  "element_list": [
    {
      "element_id": 835266081714884677
    }
  ]
}

3.5 Kling Virtual Try-On Model

Public model:

• kolors-virtual-try-on-v1
• kolors-virtual-try-on-v1-5

Minimum input parameters:

• model + human_image + cloth_image

Description:

• These two models are already complete primary models themselves, so there is no need to pass model_name again.

Example:

{
  "model": "kolors-virtual-try-on-v1",
  "human_image": "https://example.com/human.jpg",
  "cloth_image": "https://example.com/cloth.jpg"
}