Skip to content

Latest commit

 

History

History
117 lines (85 loc) · 3.97 KB

Camera snapshot.md

File metadata and controls

117 lines (85 loc) · 3.97 KB

Camera Snapshot Tutorial

This tutorial explains how to capture snapshots from cameras in the EVA system. There are two main approaches, each with different advantages and use cases.

Overview

The EVA system supports two methods for taking camera snapshots:

  1. WebRTC Stream Snapshots - Fast, simultaneous capture from compressed streams
  2. Direct Camera Snapshots - Higher quality capture directly from camera hardware

Method 1: WebRTC Stream Snapshots

Description

This method captures snapshots from the WebRTC streaming feed. While the image quality is lower due to compression, it allows for instantaneous capture from multiple cameras simultaneously.

Use Cases

  • Stereo camera operations where simultaneity of both photos is essential
  • Real-time applications requiring immediate capture
  • When network bandwidth is limited

How it Works

  • Connects to Janus WebRTC gateway
  • Receives compressed video streams
  • Captures frames directly from the stream
  • Processes multiple streams simultaneously

Implementation

A Flask server runs locally on the computer and:

  • Connects to the Janus WebRTC server
  • Receives all camera streaming feeds
  • Provides API endpoints for snapshot capture

Repository: [WebRTC snapshot repo]

Advantages

  • ✅ Instantaneous capture
  • ✅ Simultaneous multi-camera snapshots
  • ✅ No interruption to streaming

Disadvantages

  • ❌ Lower image quality (compressed stream)
  • ❌ Limited by streaming resolution

Method 2: Direct Camera Snapshots

Description

This method captures photos directly from the camera hardware, providing higher quality images at the cost of sequential processing and potential streaming interruption.

Use Cases

  • High-quality image capture requirements
  • Documentation and detailed analysis
  • When image quality is more important than speed

How it Works

The snapshot server (Flask-based) runs on each Raspberry Pi and directly accesses the camera hardware.

DWE Cameras

  • Expose a secondary MJPEG device for snapshots
  • H.264 video streaming continues uninterrupted
  • Multiple photos are taken sequentially due to Raspberry Pi limitations

Other MJPEG Cameras

  • Streaming process is temporarily stopped
  • High-quality photo is captured
  • Streaming resumes after capture
  • Causes brief lag in NEXUS during photo capture

Implementation

Repository: [snapshot server repo]

Advantages

  • ✅ Higher image quality
  • ✅ Direct hardware access
  • ✅ Full camera resolution available

Disadvantages

  • ❌ Sequential capture only (Raspberry Pi limitation)
  • ❌ Potential streaming interruption (MJPEG cameras)
  • ❌ Brief lag in NEXUS interface

Choosing the Right Method

Requirement WebRTC Snapshots Direct Snapshots
High image quality ❌ ✅
Simultaneous capture ✅ ❌
No streaming interruption ✅ Depends on camera
Real-time applications ✅ ❌
Stereo vision ✅ ❌

Technical Architecture

WebRTC Method

Computer → Janus Gateway → Camera Streams → Flask Server → Snapshot API

Direct Method

Raspberry Pi → Camera Hardware → Flask Server → Snapshot API

Getting Started

  1. For WebRTC snapshots: Set up the local Flask server and connect to Janus
  2. For direct snapshots: Deploy the snapshot server to each Raspberry Pi
  3. Choose the appropriate method based on your quality vs. speed requirements

Repository Links

  • WebRTC Snapshot Server: [repo]
  • Direct Snapshot Server: [repo]

Note: The choice between methods depends on your specific use case. For stereo vision applications, WebRTC snapshots are essential due to the simultaneity requirement. For single high-quality captures, direct camera snapshots provide superior image quality.