Data output

Session folder

RSLogger creates one folder per session inside the folder you choose when you click Start. Each module writes into its own subfolder. A control file at the top level records every button press.

session_YYYYMMDD_HHMMSS/
├── YYYYMMDD_HHMMSS_CONTROL.csv
├── Cameras/
│   └── <camera id>/
│       ├── <token>_CAM_trial<NNN>_<camera>.mp4
│       └── <token>_CAM_trial<NNN>_<camera>_timing.csv
├── picam0/                                       (CSI camera, Raspberry Pi)
│   ├── <token>_CSI_trial<NNN>_<camera>.avi
│   └── <token>_CSI_trial<NNN>_<camera>_timing.csv
├── Audio/
│   ├── <token>_AUD_trial<NNN>_MIC<n>_<microphone>.wav
│   └── <token>_AUD_trial<NNN>_MIC<n>_<microphone>_timing.csv
├── EyeTracker-Neon/
│   ├── <token>_ET_trial<NNN>_WORLD_<w>x<h>_<fps>fps.mp4
│   ├── <token>_ET_trial<NNN>_EYES_384x192_<fps>fps.mp4
│   ├── <token>_ET_trial<NNN>_AUDIO.wav
│   ├── <token>_ET_trial<NNN>_GAZE.csv
│   ├── <token>_ET_trial<NNN>_EVENTS.csv
│   └── <token>_ET_trial<NNN>_IMU.csv
├── DRT/
│   └── <token>_DRT_<device>.csv
├── VOG/
│   └── <token>_VOG_<port>.csv
├── GPS/
│   └── <token>_GPS_<device>.csv
└── Notes/
  └── <token>_NTS_trial<NNN>_notes.csv

File names

Cameras, audio, the eye tracker and notes write a new file set for each trial. The DRT, the goggles and GPS keep one file per device open for the whole session and append a row for each event, so the trial is read from the trial column.

Columns shared by every module

Every CSV that a module writes starts with the same six columns.

Control file

YYYYMMDD_HHMMSS_CONTROL.csv sits in the session folder and has one row per event in the main window.

This file is the record of when you pressed Record and Pause, and it is the simplest way to find the start of a trial on the shared clock.

Cameras

One MP4 and one timing CSV per camera per trial. USB cameras write H.264 in an MP4 container, with AAC audio when the camera’s microphone is enabled. CSI cameras on the Raspberry Pi write MJPEG in an AVI container.

Timing CSV, 9 columns

For a USB camera, record_time_* is taken when the frame is read from the camera driver.

Audio

One WAV and one timing CSV per microphone per trial. The WAV is 16-bit PCM, one channel, at the sample rate in the module settings. A microphone with several channels is recorded from its first channel.

Timing CSV, 8 columns

Each row is one chunk of samples as the driver delivered it. record_time_* is taken in the driver callback when the chunk arrives.

Eye tracker

Per trial: scene video, eye video, optional scene audio, and three CSV files. The scene video is H.264 in MP4 at the resolution and frame rate set in the module. The eye video is 384 × 192 pixels, both eyes side by side.

GAZE CSV, 36 columns

EVENTS CSV, 24 columns

IMU CSV, 19 columns

Units for the IMU, event and eye-state fields are those of the Pupil Labs real-time API, which the Neon documentation defines.

Detection response task

One CSV per DRT device for the whole session. Each row is one stimulus.

Wired DRT, 11 columns

Wireless DRT, 11 columns

The same columns as the wired unit through reaction_time_ms, except that device_time_unix carries the device’s real-time clock, then:

RSLogger logs the reaction time as the device reports it. It does not apply the 100–2500 ms window that ISO 17488 uses to classify hits. The DRT page shows how to apply it in analysis.

Visual occlusion goggles

One CSV per goggle controller for the whole session. Each row is one trial.

Wired goggles, 9 columns

Wireless goggles, 12 columns

The same nine columns, with device_time_unix carrying the device’s real-time clock, then:

GPS

One CSV per receiver for the whole session. Each row is one parsed NMEA sentence.

GPS CSV, 25 columns

RSLogger keeps a running snapshot of the fix, updates it from each sentence and writes the whole snapshot per row, so each row holds the latest known value of every field. A field is empty until a sentence that carries it has arrived.

Notes

One CSV per trial. Each row is one note.

Notes CSV, 8 columns

Timestamp formats

Six decimals on a wall-clock value and nine on a monotonic value describe how the numbers are written, not how accurately any event was timed. Synchronization explains what each timestamp does and does not tell you.