Bedrock Nova

English | 中文

A comprehensive collection of examples demonstrating the capabilities of Amazon Bedrock Nova models, including image understanding, image creation, video understanding, video creation, and text generation.

Features

• 🖼️ Image Processing:
  • Image understanding: analysis, Q&A, classification, and summarization
  • Image creation: generate images from text descriptions
• 🎥 Video Processing:
  • Video understanding: analysis, Q&A, and content summarization
  • Video creation: generate videos from text or images
• 📝 Text Generation: Both streaming and non-streaming text generation capabilities
• 🔢 Multimodal Embeddings (MME): Generate embeddings for text and images for semantic search and similarity comparison

Prerequisites

• AWS Account with Bedrock access
• Python 3.7+
• Boto3
• AWS credentials configured

Installation

1. Clone the repository:

git clone https://github.com/yourusername/bedrock-nova.git
cd bedrock-nova

2. Install dependencies:

pip install boto3

3. Configure AWS credentials:

aws configure

Usage

Image Understanding

python3 images/nova_image_understanding.py

Image Creation

python3 images/nova_image_creation.py

Video Understanding

python3 video/nova_video_understanding.py

Video Creation

# Create video from image
python3 video/nova_video_creation_by_image.py

# Create video from text
python3 video/nova_video_creation.py

Text Generation

# Non-streaming text generation
python3 text/nova_text_generation.py

# Streaming text generation
python3 text/nova_text_generation_streaming.py

Multimodal Embeddings (MME)

# Text embedding
python3 mme/nova_mme_demo.py

# Image embedding
python3 mme/nova_image_embedding_demo.py

Technical Specifications

Image Understanding

• Total payload size limit: 25MB
• Supported aspect ratios: 1:1 to 1:9, 2:3, 2:4, and their transposes
• Minimum dimension: At least one side > 896px
• Maximum resolution: 8000x8000 pixels

Image-to-Token Conversion

Image Resolution	Estimated Tokens
900 x 450	~800
900 x 900	~1300
1400 x 900	~1800
1800 x 900	~2400
1300 x 1300	~2600

Video Understanding

• Single video per payload
• Base64 payload limit: 25MB
• S3 URI video size limit: 1GB
• Supported formats: MP4, MOV, MKV, WebM, FLV, MPEG, MPG, WMV, 3GP

Video Processing

• Resolution: All videos converted to 672x672 square
• Frame sampling:
  • ≤16 minutes: 1 FPS

  • 16 minutes: Reduced rate, fixed 960 frames

• Recommended durations:
  • Low motion: <1 hour
  • High motion: <16 minutes

Token Usage

Video Duration	Frames Sampled	Sample Rate (FPS)	Token Count
10 sec	10	1	2,880
30 sec	30	1	8,640
16 min	960	1	276,480
20 min	960	0.755	276,480
30 min	960	0.5	276,480
45 min	960	0.35556	276,480
1 hr	960	0.14	276,480
1.5 hr	960	0.096	276,480

Project Structure

bedrock-nova/
├── images/
│   ├── nova_image_creation.py
│   ├── nova_image_understanding.py
│   └── test1.png
├── mme/
│   ├── nova_mme_demo.py
│   └── nova_image_embedding_demo.py
├── text/
│   ├── nova_text_generation.py
│   └── nova_text_generation_streaming.py
├── video/
│   ├── nova_video_creation.py
│   ├── nova_video_creation_by_image.py
│   ├── nova_video_understanding.py
│   └── animals.mp4
├── README.md
└── README_zh.md

Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.