Good captures produce good data. These practices reduce the most common sources of error.
Environment
- Use a consistent space. Capturing the same subject in the same space across sessions minimizes confounding variables.
- Neutral background. Avoid busy patterns, mirrors, or moving objects behind the subject.
- Adequate floor space. Allow enough room for the full movement (at least 4–5 meters for gait).
- No slippery surfaces during dynamic tests.
Lighting
- Even, diffuse lighting. Overhead fluorescent or natural diffuse light works well.
- Avoid backlighting. A window or lamp directly behind the subject creates silhouettes and defeats pose estimation.
- Consistent across sessions. Changes in lighting affect joint detection accuracy.
Clothing
- Contrast with the background. If the floor is dark, avoid black clothing.
- Tight to body for joint identification. Loose clothing obscures the hip, knee, and shoulder centers.
- Avoid shiny fabrics that confuse pose estimation models.
- Shoes. Use the subject's normal training or walking shoes. Barefoot is also acceptable if consistent across sessions.
Camera placement
- 45° angle pair. The two cameras should be at roughly 45° offset from the direction of motion.
- Waist height. Mount cameras at approximately the subject's waist height to minimize perspective distortion.
- Distance. 3–4 meters from the subject gives a good balance between full-body visibility and spatial resolution.
Subject coaching
- Coach the subject through the movement exactly as you would without the platform.
- QOOM measures what the subject actually does — if you over-coach or under-coach, the results reflect that.
- Record at least one warm-up trial that you discard.
Session hygiene
- Always set a consistent subject name or external ID.
- Add session notes (planned in a future release).
- Recalibrate at the start of each session, even if the cameras have not obviously moved.