AWS Bedrock - 圖像生成

使用 Bedrock 搭配 Stable Diffusion、Amazon Titan Image Generator 和 Amazon Nova Canvas 模型進行圖像生成。

支援的模型

模型名稱	函式呼叫	成本追蹤
Stable Diffusion 3 - v0	image_generation(model="bedrock/stability.stability.sd3-large-v1:0", prompt=prompt)	✅
Stable Diffusion - v0	image_generation(model="bedrock/stability.stable-diffusion-xl-v0", prompt=prompt)	✅
Stable Diffusion - v1	image_generation(model="bedrock/stability.stable-diffusion-xl-v1", prompt=prompt)	✅
Amazon Titan Image Generator - v1	image_generation(model="bedrock/amazon.titan-image-generator-v1", prompt=prompt)	✅
Amazon Titan Image Generator - v2	image_generation(model="bedrock/amazon.titan-image-generator-v2:0", prompt=prompt)	✅
Amazon Nova Canvas - v1	image_generation(model="bedrock/amazon.nova-canvas-v1:0", prompt=prompt)	✅

用法

SDK

基本用法

import os
from litellm import image_generation

os.environ["AWS_ACCESS_KEY_ID"] = ""
os.environ["AWS_SECRET_ACCESS_KEY"] = ""
os.environ["AWS_REGION_NAME"] = ""

response = image_generation(
    prompt="A cute baby sea otter",
    model="bedrock/stability.stable-diffusion-xl-v0",
)
print(f"response: {response}")

設定選用參數

import os
from litellm import image_generation

os.environ["AWS_ACCESS_KEY_ID"] = ""
os.environ["AWS_SECRET_ACCESS_KEY"] = ""
os.environ["AWS_REGION_NAME"] = ""

response = image_generation(
    prompt="A cute baby sea otter",
    model="bedrock/stability.stable-diffusion-xl-v0",
    ### OPENAI-COMPATIBLE ###
    size="128x512", # width=128, height=512
    ### PROVIDER-SPECIFIC ### see `AmazonStabilityConfig` in bedrock.py for all params
    seed=30
)
print(f"response: {response}")

PROXY

1. 設定 config.yaml

model_list:
  - model_name: amazon.nova-canvas-v1:0
    litellm_params:
      model: bedrock/amazon.nova-canvas-v1:0
      aws_region_name: "us-east-1"
      aws_secret_access_key: my-key # OPTIONAL - all boto3 auth params supported
      aws_secret_access_id: my-id # OPTIONAL - all boto3 auth params supported

2. 啟動 proxy

litellm --config /path/to/config.yaml

3. 測試它！

文字轉圖像：

curl -L -X POST 'http://0.0.0.0:4000/v1/images/generations' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer $LITELLM_VIRTUAL_KEY' \
-d '{
    "model": "amazon.nova-canvas-v1:0",
    "prompt": "A cute baby sea otter"
}'

色彩引導生成：

curl -L -X POST 'http://0.0.0.0:4000/v1/images/generations' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer $LITELLM_VIRTUAL_KEY' \
-d '{
    "model": "amazon.nova-canvas-v1:0",
    "prompt": "A cute baby sea otter",
    "taskType": "COLOR_GUIDED_GENERATION",
    "colorGuidedGenerationParams":{"colors":["#FFFFFF"]}
}'

Amazon Nova Canvas - 圖像編輯

搭配 Bedrock Nova Canvas（amazon.nova-canvas-v1:0）使用與 OpenAI 相容的 image_edit()。請求使用與生成相同的 InvokeModel API；LiteLLM 會將輸入映射到 Nova Canvas task types：

情境	傳送至 Bedrock 的 taskType
圖像 + 提示詞（無遮罩）	IMAGE_VARIATION
圖像 + 提示詞 + 遮罩	INPAINTING（inPaintingParams.image、maskImage 或 maskPrompt）
taskType: OUTPAINTING + mask 或 maskPrompt	OUTPAINTING（Bedrock 需要其中一個；如果兩者都缺少，LiteLLM 會清楚地報錯）
taskType: BACKGROUND_REMOVAL	BACKGROUND_REMOVAL

from litellm import image_edit

response = image_edit(
    image=open("photo.png", "rb"),
    prompt="Add soft sunset lighting",
    model="bedrock/amazon.nova-canvas-v1:0",
)

對於 BACKGROUND_REMOVAL，AWS 請求不得包含 imageGenerationConfig；即使您傳入 size、n、seed 等，LiteLLM 也會在該任務中省略它。用於圖像編輯的其他 Nova Canvas inference IDs 應在 model_prices_and_context_window.json 中設定 supports_nova_canvas_image_edit: true（請參閱 amazon.nova-canvas-v1:0）。

使用圖像生成的 Inference Profiles

對於帶有圖像生成功能的 AWS Bedrock Application Inference Profiles，請使用 model_id 參數來指定 inference profile ARN：

SDK

from litellm import image_generation

response = image_generation(
    model="bedrock/amazon.nova-canvas-v1:0",
    model_id="arn:aws:bedrock:eu-west-1:000000000000:application-inference-profile/a0a0a0a0a0a0",
    prompt="A cute baby sea otter"
)
print(f"response: {response}")

PROXY

model_list:
  - model_name: nova-canvas-inference-profile
    litellm_params:
      model: bedrock/amazon.nova-canvas-v1:0
      model_id: arn:aws:bedrock:eu-west-1:000000000000:application-inference-profile/a0a0a0a0a0a0
      aws_region_name: "eu-west-1"

驗證

圖像生成支援所有標準 Bedrock 驗證方法。詳情請參閱 Bedrock Authentication。