leafjiang / vimbagrabber

vimbaGrabber

Simple grabber recording synchronized videos from all connected Allied Vision Technologies USB 3.0 cameras using the Vimba SDK.

Building on Ubuntu 16.04

0. Install build essentials: sudo apt-get install build-essential.
1. Install video encoder and opencv dependencies: sudo apt-get install libavcodec-dev libswscale-dev libavdevice-dev libavfilter-dev libavformat-dev opencv.
2. Install the VimbaSDK. IMPORTANT: Make sure to install both the GigE and USB3 drivers or the USB 3.0 camera discovery might be broken. Note for C APIs: a call of DiscoverGigECameras() before CamerasList() is needed to find USB cameras.
3. Modify VIMBASDK_DIR in build/Makefile to point to the SDK's extraction folder. Alternatively you can create a symbolic link to the extracted SDK archive in this projects root folder: ln -s /path/to/Vimba_2_0 /path/to/vimbaGrabber.
4. Start the build with cd build; make. The executable will be in build/binary/<architecture>/vimbaGrabber.

Installation:

To make vimbaGrabber accessible across the system simply generate a symbolic link inside /usr/local/bin:

sudo ln -s /path/to/vimbaGrabber/build/binary/<architecture>/vimbaGrabber /usr/local/bin/vimbaGrabber

Usage:

Usage: vimbaGrabber ([options]) 
  Output synchronized videos from every connected vimba camera.
  The output files will be placed in the current folder and 
  named "<output label>_runXX_cameraXX.h264".

Options:
  -w , --width :    Set the cameras and output videos to this width
                                   default: 640
  -h , --height : Set the cameras and output videos to this height
                                   default: 480
  -f , --fps :          Set the cameras and output videos to this fps
                                   default: 24
  -c , --crf :          Set the h246's constant rate factor, the higher the lossier
                                   default: 0 (lossless)
  -s, --silent:                   Do not display the live video while recording
                                  

The output files are the raw lossless (crf=0) h264 encoded video streams. ###What to do with the output files: In case your program can't read the raw frame data you need to pack it into a video container using ffmpeg. For example in order to pack a proper Matroska (*.mkv) container without reencoding run:

ffmpeg -i testrecording_run01_camera01.h246 -codec copy testrecording_run01_camera01.mkv

You can also extract the single (uncompressed) frames as follows:

ffmpeg -i testrecording_run01_camera01.h246 testrecording_run01_camera01frame%04d.png