Why tank size changes what calibration has to solve
A trajectory tracking system does not see tanks. It sees pixels. The job of calibration is to turn those pixels into real-world coordinates so that a distance moved in the arena can be reported in millimeters and a speed can be reported in millimeters per second. That conversion sounds mechanical, but the tank size changes which problems matter most. A small 24-well plate has a narrow field of view and shallow depth. A large adult tank has a wide field of view and much more depth. The calibration approach has to adapt to both.
The core issue is scale. A tracking system that reports distance in pixels is not useful for comparing behavior across tanks. A larva in a small well might move a few hundred pixels, while an adult in a large tank might move a few thousand. Without a consistent physical scale, those numbers cannot be compared. Calibration provides that scale.
The basics of pixel to millimeter conversion
Every calibration starts with a known distance. A ruler, a grid, or a printed target with a known dimension is placed in the tank or arena. The camera captures an image, and the software identifies the reference marks. The number of pixels between two marks is divided by the known distance to produce a scale factor, usually expressed as millimeters per pixel.
That scale factor is then applied to every measurement. A movement of 100 pixels becomes a movement of 100 times the scale factor in millimeters. The math is simple. The complications come from the tank itself.
-
The reference target must lie in the same plane as the fish.
-
The camera must not move between calibration and recording.
-
The scale factor must be stored with the recording settings.
-
Any change in lens, zoom, or camera height invalidates the old calibration.
Depth, perspective, and why large tanks are harder
A small well plate is nearly flat. The fish swims in a shallow layer of water, and the camera looks down from above. The plane of the fish and the plane of the calibration target are almost the same. A large tank has vertical structure. Fish can swim at the surface, in the middle, or near the bottom. A camera looking down from above sees the fish at different apparent sizes depending on depth. A fish near the bottom appears slightly larger or smaller than one at the surface, depending on the lens and the camera height.
This is a perspective problem. It can be reduced by using a telecentric lens or by keeping the camera far from the tank. It can also be corrected in software if the camera height and lens parameters are known. For many behavioral studies, the correction is small enough to ignore. For studies that compare fine differences in speed or distance, the correction matters.
| Tank type | Typical depth range | Perspective impact | Calibration approach |
|---|---|---|---|
| 24-well plate | Shallow, near flat | Minimal | Single-plane calibration |
| 6-well plate | Shallow, near flat | Low | Single-plane calibration |
| Small larval tank | Moderate | Low to moderate | Single-plane with depth note |
| Adult tank | Deep | Moderate to high | Multi-plane or depth correction |
| Large arena | Deep and wide | High | Depth-aware calibration |
Matching camera field of view to tank dimensions
The field of view has to cover the entire arena. If the tank extends beyond the camera frame, fish that swim near the edge will be lost. That sounds obvious, but it becomes a real constraint when a facility uses one camera system across tanks of different sizes.
A camera set up for a small plate will have a narrow field of view. Moving it to a large tank requires either repositioning the camera further away or switching to a wider lens. Both changes invalidate the calibration. A facility that works with multiple tank sizes needs a calibration routine that can be repeated quickly and stored per configuration. Some systems support multiple calibration profiles that can be loaded when the tank type changes.
Recalibration triggers and documentation
Calibration is not a one-time event. It has to be repeated whenever something in the optical path changes. Camera movement, lens changes, and even a bump to the rig can shift the mapping. A tracking system that does not track these changes will produce data that looks fine but is subtly wrong.
-
Recalibrate after any camera or lens adjustment.
-
Recalibrate after moving the tank or the rig.
-
Verify calibration with a known distance at the start of each session.
-
Document the calibration date and settings with the experiment record.
A quick verification step is worth the time. Placing a ruler in the arena and checking that the software reports the correct length takes seconds. If the reported length is off, the calibration is off. That check catches the most common failure modes before they affect an experiment.
A field example from a facility running mixed tank sizes
A behavioral core facility in eastern China supported studies ranging from 96-well plate larval assays to large adult tanks. The original workflow used a single camera setup and recalibrated manually whenever the tank changed. The manual process was slow and error-prone. Technicians sometimes forgot to recalibrate, and the resulting data showed impossible speeds.
The team reorganized the workflow. Each tank type was assigned a calibration profile with a stored scale factor and camera height. A printed reference target was kept with each tank type. Before each session, the operator loaded the correct profile and verified it with a quick ruler check. The error rate dropped. The setup time also dropped, because the profile loaded faster than a full manual recalibration.
The change was not a new camera or a new algorithm. It was a workflow change that matched the calibration process to the reality of mixed tank sizes.
Where calibration gets fuzzy and what to accept
No calibration is perfect. Lens distortion, water refraction, and depth variation all introduce small errors. For many behavioral endpoints, those errors are small enough to ignore. For studies that measure fine differences in velocity or path curvature, they may not be.
The honest approach is to know the size of the error and to report it. A calibration check with a known distance gives an estimate of the residual error. If that error is small relative to the biological effect being studied, the calibration is good enough. If it is large, the setup needs adjustment.
It also helps to keep the camera and tank configuration stable within a study. Changing the setup mid-study introduces a calibration shift that can be mistaken for a treatment effect. Consistency in hardware is part of consistency in data.
Hunter Lab (Hunter) builds zebrafish laboratory equipment with attention to the optical stability that trajectory tracking depends on. The manufacturing and quality control side supports the consistency that multi-tank studies require. For labs calibrating systems across different tank sizes, that kind of build discipline is part of the foundation.