This repository contains the VISSS data acquisition software. Please see also
- VISSS data processing library https://github.com/maahn/VISSSlib
- VISSS2 hardware plans https://zenodo.org/doi/10.5281/zenodo.7640820
- VISSS3 hardware plans https://zenodo.org/doi/10.5281/zenodo.10526898
- original version for MOSAiC
- running on (too slow) Intel i7-4770
- capture_id overflows at 65535 - relatively easy to fix
- capture_time drifts because it is only reset when camera starts
- record_time is assigned in the processing queue, can be a couple of seconds off if queue is long
- flipped capture_time: once in a while timestamps of two consecutive frames are flipped. Not clear whether frame itself is also flipped. So far only observed for follower. Solution: remove flipped frames assuming meta data and frame are flipped.
- ghost frames: sometimes a couple of frames are less than 1/fps apart. looks like an additonal frame is inserted and delta t is reduced accordingly for approximate 6 frames to make up for additional frame. Origin of additonal frame is unclear. So far only observed for follower.
- file_starttime: obtained from record_time, so problems with record_time apply
- For Hyytiälä and Ny-Ålesund deployments in 21/22
- New Python interface
- New configuration files
- Ability to remote control instrument
- New log files
- extended meta data
- file_starttime: obtained from record_time, so problems with record_time apply
- record_time is assigned in the processing queue, can be a couple of seconds off if queue is long
- camera time request takes a couple of frames to become active, but is applied in data acquisition immediately
- .mov file sometimes broken if it contains only a single frame
- Using ffmpeg pipe instead of opencv videowriter. Gives better thread control does not require custom patched opencv version any more.
- For deployments in 22/23 and after
- changed from mov to mkv files to handle crashes better
- write last timestamp to ascii file when closing (actually first time stamp of next file...)
- restart data acqusition when reset clock fails
- assign record_time as early as possible
- apply camera time request when it is reset in camera
- fix cature_id overflow
- crate timestamp in filename smarter - from camera reset_time (capture_time)
- C++ restart after camera failure not reported in status file leading to capture_id reset no beeing handled properly (fixed on 14 April 2023)
- For deployments in 24/25 and after
- Store internal camera temperature as status information