Radiation-free, low-cost methods are needed for preliminary scoliosis screening when repeated radiographic assessment is undesirable. This pilot validation study evaluated a Kinect-based RGB-D optical photogrammetry system for surface-based assessment of spinal deformity in 40 participants with suspected scoliosis recruited from a single clinical centre (21 female and 19 male; age range, 7–77 years; BMI range, 15.97–28.00 kg/m²; Cobb-angle distribution: 10 participants < 10°, 26 participants > = 10 and < 20°, and 4 participants > = 20 and < 40°). Kinect colour and depth images were aligned to generate a 3D point cloud. The workflow integrates filtering, colour-based segmentation, smoothing and simplified manual landmarking, landmark-assisted automated spine-line extraction, and computes shoulder asymmetry, the deep optics-based scoliosis assessment (DOSA) angle, the Sinoto kyphosis index, as well as the angle of trunk rotation. Radiographic Cobb angles were used as the reference standard for coronal-plane validation. The mean DOSA-Cobb difference was − 0.39° (95% CI, −1.28 to 0.50), with Bland-Altman 95% limits of agreement ranging from − 5.85 to 5.08°. At the 10° diagnostic threshold, sensitivity was 83.3%, specificity was 90.0%, accuracy was 85.0%, and the ROC AUC was 0.967. These findings indicate that Kinect-based RGB-D optical photogrammetry offers a low-cost screening solution by integrating automated spinal line extraction and surface metrics from multiple anatomical planes. However, larger independent cohorts, adolescent-specific validation, and formal reliability testing are required before routine clinical, school-based, community, or remote use.