Field protocol

A written field protocol connects each recording to the information needed to interpret it later. The protocol should specify how equipment is prepared, how recordings are identified and synchronized, which supporting observations are collected, and how the files and field notes are preserved. Pilot work under realistic conditions is the best time to find omissions and revise the protocol.

Before fieldwork

Define the observations and outputs. Decide which objects, events, and supporting measurements will be recorded, and write operational definitions that different observers can apply consistently.
Test the complete procedure. Make pilot recordings under conditions similar to those expected in the study. Check subject visibility, field of view, focus, motion blur, glare, obstructions, and the visibility of any reference objects used later.
Prepare and inspect the equipment. Charge batteries, clear or replace recording media, clean lenses, inspect mounts and housings, and test every cable and recorder. Underwater housings require particular attention to seals, O-rings, condensation, and focus through the housing and water.
Choose and record camera settings. The field record should include resolution, frame rate, exposure settings or mode, focus, zoom or focal length, and stabilization settings. Record settings again whenever they change.
Plan file and site identifiers. Use short, unambiguous codes that can appear on a slate or whiteboard, in field notes, and in filenames. Decide how camera identifiers and successive recording sessions will be distinguished.
Plan synchronization and supporting footage. Specify the cue that will be visible in all cameras and the footage needed for calibration, distortion correction, environmental measurements, surfaces, known objects, or tracers used by the study.
Assign responsibilities. When several people are involved, state who operates each camera, records field notes, performs the synchronization cue, collects associated measurements, and verifies the files. Any work involving animals must follow the applicable permits, approved animal-care procedures, and safety requirements.

During recording

Identify the recording. Film a slate or whiteboard showing the site, date, camera identifier, and session or deployment number. Record the corresponding filenames and clock time in the field notes.
Record a synchronization cue. Produce a brief, unambiguous event that is visible in every camera, such as a single LED flash. Repeat it after cameras are restarted or whenever the relationship among clips may have changed.
Check the setup before leaving it. Confirm that every camera is recording, the intended area is visible and in focus, settings have not shifted, housings are sealed, and mounts are secure.
Record the primary observations. Allow subjects to resume normal behavior when disturbance is possible. Note events that may affect interpretation, including people or equipment entering the scene, changes in lighting or visibility, passing boats, and interruptions in recording.
Collect the associated measurements and reference footage. Depending on the study, this may include environmental measurements, surface or depth references, tracer sequences, site photographs, known objects, calibration footage, and distortion-correction footage. Record the time and purpose of each sequence. Film the chessboard separately for each camera, rotating it so that it faces that camera approximately square-on and fills much of the frame. With convergent cameras, this usually requires a different chessboard orientation for each camera.
Document any change in the setup. If a camera or mount moves, or if focus, zoom, resolution, or another relevant setting changes, record when it happened. Footage recorded after the change may need to be treated as a separate setup during analysis.
Verify the required material before dismantling the setup. When camera geometry must remain fixed, complete the observation and associated reference footage before moving any camera.

After recording

Copy and back up the original files promptly. Preserve the camera originals and create a second copy on separate storage before cards are erased or reused.
Use consistent filenames without losing the originals. Filenames should connect the clips to the project, site, date, camera, and session recorded in the field notes.
Review representative footage. Confirm that each file opens, the subject and reference material are usable, synchronization cues are visible, and the recorded settings agree with the field notes.
Record problems while they are still fresh. Note missing clips, uncertain synchronization, camera movement, setting changes, poor visibility, or departures from the planned procedure. These notes are often more informative than trying to reconstruct the problem during analysis.
Keep the field record with the project data. Retain it through analysis and archiving so that filenames, sites, cameras, and associated measurements remain traceable.

Suggested field record

At minimum, record the following:

project, site, date, session, and personnel;
camera identifiers, filenames, recording start and stop times, and battery or media changes;
resolution, frame rate, exposure, focus, zoom or focal length, stabilization, and other study-relevant settings;
mounting arrangement, approximate subject distance, and any movement of cameras or mounts;
synchronization cues and their approximate times in each recording;
environmental conditions and associated samples or measurements;
times and purposes of calibration, distortion-correction, surface, tracer, or other reference footage;
departures from the protocol, equipment problems, disturbances, and other observations relevant to interpretation; and
locations of the original files and their backups.

Using the recordings in VidSync

Field identifiers and notes should carry directly into the VidSync project. See Managing video clips in VidSync for importing, naming, and synchronizing clips, and Measuring “Objects” and “Events” in VidSync for organizing observations and applying a consistent measurement protocol.

Historical example: juvenile Chinook salmon in the Chena River

The following study-specific protocol was used to collect footage of juvenile Chinook salmon feeding behavior in the Chena River, Alaska, during Jason Neuswanger’s Ph.D. research. Its equipment details and study measurements should be read as a historical example rather than a general VidSync procedure. It shows the level of detail that can prevent missed steps during fieldwork, especially during the first excursions and pilot work.

Evaluate site suitability. Individual protocols should list study-specific biological criteria as well as video quality criteria (e.g., lighting, field of view, background, contrast, obstructions between cameras and fish).

Verify lenses are clean of debris or residue.

Clean and re-grease the waterproof housing O-rings.

Turn the cameras on.

Write down the current time and remaining battery life if the cameras are to be deployed until batteries die.

Manually focus both cameras for the intended subject distance. A camera’s displayed focus distances aren’t correct for shooting through a housing into water; develop custom focus distances by testing beforehand.

Place cameras in housings and double-check fasteners.

Place a desiccant packet in the housings to prevent condensation-caused fogging in cold water.

Verify the zoom setting on both cameras (usually, widest angle). Keep zoom and focus fixed hereafter.

Start recording. Write down the time, so video timecodes can be linked to real times and noteworthy observations (e.g., a loud boat passing by).

Videotape a whiteboard with site name & date. This prevents confusion when naming or reorganizing video files. Site name may correspond to a GPS waypoint name.

Blink an LED light once, in view of all cameras, for synchronizing video clips in VidSync.

Videotape the checkerboard in each camera separately with both board and cameras submerged. It should face each camera as flatly as possible, and be far enough away to be in focus, but close enough to fill the screen. Avoid the uneven lighting of bright sunlight under a wavy surface, and seek to film in bright shade.

Videotape the calibration frame with similar lighting, with both frame and cameras as still as possible, making sure both faces are close enough to the cameras to take up a reasonably large portion of the screen. Only one ideal shot is needed, but try many poses to assure a good view. Allow time for settling of any disturbed detritus that might block the view of the frame nodes.

Verify that the cameras are still set properly, double-checking recording status, zoom, and focus.

Deploy the cameras to observe fish, securing the system as needed.

Individual protocols should describe additional measurements (e.g., drift net samples, temperature measurements, above-water site photographs) to be taken during filming at each video site.

Historical example: drift-feeding fishes in Alaska

This study-specific protocol was used during my postdoctoral research in Alaska through the University of Georgia, studying the behavior of three species of drift-feeding fish with VidSync. If the reference to cous cous in step 1 takes you by surprise, it is because we tested many materials and found that Israeli cous cous (a totally different substance from regular cous cous), after soaking in water, made exceptionally good, near-neutral-buoyancy, highly visible tracers for computing water velocity fields in VidSync. We used two Nikon D5300 DSLR cameras wired to Atomos Ninja HDMI recorders on shore. I would not recommend that specific setup to my worst enemy, but we collected great data. The equipment, settings, distances, exposure values, and animal-handling steps below belonged to that study and are retained as a historical example. Their broader value is in showing the kinds of details that may need to be tested and written down before fieldwork and refined during pilot work.

Start soaking Israeli cous cous and chopped marabou velocity tracers at least an hour in advance of when they’ll be used.
Find a good fish to film (or Chinook school) and remember its exact location.
1. If necessary, use the GoPro on carbon fiber rod to find feeding fish and identify their exact focal point in relation to underwater landmarks.
1. Unless filming Chinook, try to find a fish is feeding by itself with minimal visible influence from nearby competitors.
Prepare drift nets:
1. Find a representative location and pound in the rebar for it, somewhere the net(s) can be handled without scaring the fish.
1. Use the PVC pole to measure depth at the drift net site. Record it now and use it later to determine where to place the net(s) on the pole.
1. Fix the drift nets to the pole, but don’t place them yet. Write down the position of the top of each net (middle of the top tube) on the pole.
Prepare the cameras (starting outside their housings). The order of some of these steps (power / live view / connections) is very important for the cameras to run properly.
1. Fix the cameras tightly onto their crossbar and verify that the joints are tightened so they won’t wobble (but don’t fix the crossbar to the main mount yet).
1. Make sure the housing O-rings are clean.
1. Make sure the camera lenses are really, really clean. Smudges really show up with short-focused fisheye lenses.
1. Electrical tape a desiccant packet inside the right wall of the housing.
1. Connect the power and HDMI cables to the housing and generator.
1. Turn on Atomos units and connect them to the HDMI cables.
1. Power the cameras on. Settings as follows:
  1. 1080p30
  1. Movie quality: High
  1. Manual movie settings: On
  1. HDMI: Output resolution 1080p, device control Off
  1. If using Samyang lenses with ‘A’ mode, set aperture to f/22, which gives the camera electronic control of the real aperture.
  1. Lens and focal length strategy:
    1. Chena – Try to place the cameras 20-35 cm from the fish, with the 8 mm Samyang lenses focused at 1’. If required subject distances are 40+ cm, use the Tokina at 17 mm. Try to avoid distances of 60+ cm, although 60-80 might be okay with excellent lighting and backdrop.
    1. Panguingue – When bathymetry and focal points allow, try for subject distances in the 30-60 cm range with the Samyang lenses. For 60-100, use 17 mm, or maybe 13 on the low end of the range if there are field-of-view concerns. Try to avoid distances of 100+ cm, although 100-130 should be workable at 17 mm.
    1. Clearwater – For fish with very predictable focal points, try for subject distances in the 40-80 cm range with the Samyang lenses. In the 80-120 range, use the Tokina at 13 mm. Beyond that, use 17 mm. Try to avoid distances of 200+ cm, although 200-250 is workable at 17 mm.
    1. All sites – Use electrical tape to secure the focus and focal length rings so they can’t accidentally slide.
  1. Focus strategy:
    1. If using the Samyang 8 mm, focus at 1.2’ for Dollies and Grayling, or 1’ for Chinook.
    1. Always focus the Tokina lens on the “1” in “1.25,” regardless of focal length.
  1. Exposure strategy:
    1. Inconsistent lighting, such as thick passing clouds or deep, moving tree shadows: Use ‘A’ mode, f/11. Otherwise, use manual exposure as described below.
    1. Consistent, direct sunlight: f/11, 1/500 s, ISO 1250
    1. Consistent, brightly overcast: f/11, 1/320 s, ISO 1600
    1. Consistent, shaded from direct sun but 50%+ bright skylight: f/11, 1/250 s, ISO 1600
    1. Consistent, heavy shadows: f/8, 1/60 s, ISO 6400
    1. In general, be willing to go up to ISO 6400 before dropping shutter speed below 1/200 s. But be willing to drop down to (and not below) 1/200 s before taking ISO above 3200.
    1. Remember aperture cannot be set while live view is on. Toggle it off, set the aperture, then toggle it back on.
1. Electrical tape the focus (and zoom, for Tokina) sliders in place.
1. Turn on live view.
1. Verify that the cameras are sending the HDMI feed plain video without a bunch of stats; cycle the ‘Info’ button to fix if necessary.
1. Plug in the red override cable stub.
1. Seal the camera housing, being careful of the cables and making sure the bottom one is in its holder.
1. Fix the crossbar to the main mount using a 9/16” wrench and lock washer.
Position the cameras to record the fish.
Make sure the fish come back into position and the camera’s pointed well. If manually exposing, double check exposure and adjust shutter speed with housing knobs if needed.
Press record on both Atomos Ninjas at the exact same time (makes syncing easier).
Write down the time the cameras started recording.
After the fish have settled into normal feeding, place the drift nets:
1. Drop them over the pole.
1. Record the time they were dropped in.
1. Record the velocity of the water entering each net, in the horizontal and vertical center of either the left or the right half of the net.
From this point on, avoid walking around at all upstream of the filming site / nets / fish so as to avoid artificially raising the amount of debris in the water.
Leave cameras alone for about 2 hours to record the fish. Based on the 2015 videos we had processed by May 10, 2016, hitting our target of 100 foraging attempts per fish observed took anywhere from 4 to 8 minutes (Chinook), from 6 to 30 minutes (Dollies), and from 17 to 51 minutes (Grayling).
Pull and process the drift nets before they begin to clog and back up significantly (typically about 30 minutes, dependent on velocity/debris).
1. Record the time and velocity as before.
1. Label Whirl-paks with the site code, YYYY-MM-DD-N_site on the outside in permanent marker and in pencil on a piece of paper inside the bag.
When finished, record post-observation water temperature and time.
Try to catch the main fish we clearly observed on camera, for gastric lavage. While the fish are being caught:
1. Write down the time each fish was caught (Panguingue/Clearwater), or the single time the trap was pulled or netting was done (Chena). If the fish was definitely caught off-camera, write ‘oc’ next to the time.
1. Store the fish together in a safe container.
1. Label Whirl-paks for each fish, both on the outside in sharpie and in pencil on a piece of paper inside the bag.
1. Labels use the form “YYYY-MM-DD-N_site Fish N_fish”
Before processing fish, finish the video site and calibration data on-camera. Do not move the cameras at all until if/when it’s necessary for calibration:
1. Get surface/depth data:
  1. If the surface is visible on camera, poke a twig down through the surface (or a branch full of twigs) throughout the visible part of the surface. Record the “surface-point” time.
  1. If the surface isn’t visible, or only barely so off in the distance, measure the depth at the fish location with a drift net PVC pole. Take the measurement, then use the bubble level to get the pole perfectly vertical. Right when it’s perfectly vertical, yank it up fast. Record “vertical-pole” time.
1. Throw the cous cous and/or marabou velocity tracers such that they drift throughout the area the fish was using. Throwing cous cous far upstream helps filter out the less neutrally buoyant pieces. Try to release a small number (5-15) of particles at a time, and reach in or use a pipe to release some down near the bottom.
1. Do the 3-D (checkered dot calibration), with minimal or no movement of the cameras from the observation position if possible.
1. Do the checkerboard calibration for distortion.
For each fish captured:
1. Anesthetize it with Aqui-S (0.5 to 1 ml, i.e. cm³, per gallon H₂0). That equals about 1 mm in the bottom of the cap of our bottle, per gallon.
1. Perform gastric lavage with the syringe (Chinook) or sprayer (Grayling/Dollies). Flush stomach contents directly into Whirl-paks (Chinook), into the drift net processing screen (Dollies), or into the big screen (Grayling), and transfer to Whirl-paks from there.
1. Photograph the whiteboard with the fish code.
1. Photograph it from both sides in a Lexan holding tube to measure length and match samples with fish on video. Find uniform lighting, and keep the face of the photo tank clean of water droplets and fog.