3D Facial Views Shift Perceived Severity of Orthodontic Cases Compared With 2D Photos, With Mixed Effects

Orthodontic assessments of facial profile attractiveness have traditionally relied on two-dimensional representations, including standardized photographs, which may shape clinical judgments differently than a rotating three-dimensional model. In a cross-sectional survey study published in Digital and Aligner Orthodontics, researchers at the University of Florida report that switching evaluators from two-dimensional images to animated 3D facial models significantly changed how attractive faces appeared, how much orthodontic treatment people thought those faces needed, and whether they recommended surgery or braces alone. The differences were not uniform: some digitally altered profiles were judged more attractive and less severe in 3D, while others fared worse, suggesting that the dimensionality of an image is not a neutral technical choice but an active variable in esthetic assessment.

The team, led by Mitali Chitre Carlisle with collaborators including Sarah Abu Arqub, constructed a tightly controlled experimental system to isolate the effect of image format. They recruited two adult volunteers, one man and one woman, both of Caucasian descent, both with confirmed skeletal Class I facial relationships on cephalometric radiographs, and neither with a history of orthodontic treatment or facial plastic surgery. Standardized frontal and profile photographs were taken in natural head position with relaxed, closed lips, so that dental appearance could not sway the evaluators. The images were then imported into FaceGen software, which reconstructs a 3D facial model from 2D frontal and profile inputs, allowing the researchers to digitally reshape the lower face at defined soft-tissue landmarks.

The modifications followed published cephalometric norms. The lower anterior facial height to total anterior facial height ratio was set at the Eastman normative value of 55 percent, then shifted by four standard deviations in either direction to produce ratios of 47 percent and 63 percent, representing decreased, average, and increased vertical proportions. Independently, the positions of the maxilla and mandible were altered in 4-millimeter increments at the A and B points: compressing point B posteriorly created a convex Class II profile, while advancing it anteriorly created a concave Class III profile. Crossing three sagittal categories with three vertical categories yielded nine altered faces per subject, eighteen depictions in total. Nonessential features such as hair were removed to prevent distractions, and all other facial attributes were held constant.

Each altered face was presented in two formats. In the 2D survey, evaluators saw a static frontal view and a static profile view. In the 3D survey, they saw a single continuously rotating animation built from 190 screenshots of the facial model, sweeping 180 degrees from right lateral to left lateral view and repeating, viewable as many times as the evaluator wished. Two anonymous REDCap surveys were distributed to a university-based population of students, dentists, dental specialists, and laypeople aged 18 or older, with the 3D survey administered six months after the 2D survey to reduce recall bias. Four hundred evaluators participated, 200 per format, each rating attractiveness on a 100-millimeter visual analog scale, judging treatment need on a five-point ordinal scale adapted from the Index of Orthodontic Treatment Need, and, where any need was perceived, choosing between orthodontic treatment alone and combined orthodontic-surgical treatment.

The statistical picture after adjustment for age, gender, and profession was strikingly asymmetric between the sexes. Among female profiles, seven of the nine comparisons reached significance at adjusted P < 0.001, and every one of those seven looked more attractive in 3D: the Class II average, Class II decreased, Class II increased, Class III average, Class III decreased, and Class III increased profiles, plus the Class I increased vertical face, all gained VAS points in the rotating format. Only the Class I average and Class I decreased female profiles, the most balanced faces in the set, showed no significant difference, suggesting that harmonious proportions are judged consistently regardless of how they are displayed.

Male profiles told a more complicated story. The Class I decreased vertical face (rated 47.2 in 3D versus 36.4 in 2D) and the Class III increased vertical face (39.1 versus 27.4) were judged significantly less attractive in 3D, while the Class II increased vertical face (36.7 versus 42.6) and the Class III decreased vertical face (26.1 versus 41.1) were judged significantly more attractive. The authors interpret this divergence through sexual dimorphism in craniofacial morphology: male faces tend to show stronger mandibular projection and greater facial depth, which can accentuate concavity when depth cues are added, whereas female faces typically present smoother soft-tissue contours that soften vertical discrepancies. In other words, adding depth does not simply flatter a face; it can either mitigate or magnify an apparent skeletal discrepancy depending on the underlying pattern.

The clinical consequences were substantial. Perceived treatment need generally shifted downward in 3D, with more “no need” ratings and fewer “great” or “very great need” ratings, particularly for Class II and Class III patterns. Treatment modality selections moved in parallel. Combined orthodontic-surgical recommendations declined in 3D for most Class II and Class III profiles, for example falling from 82 percent to 54 percent for the Class III increased vertical female face and from 69 percent to 29 percent for the Class III decreased vertical male face, with corresponding increases in orthodontics-only or no-treatment responses.

Yet the direction of change was not uniformly toward less treatment. Two male profiles moved the other way: the Class I decreased vertical face saw great or very great need ratings climb from 6.2 percent in 2D to 24 percent in 3D, and its surgical recommendation rate rose from 9.5 percent to 28 percent, while the Class III increased vertical male face saw surgical recommendations rise from 33 percent to 55 percent. The authors suggest that when facial depth and contour are fully visualized, certain skeletal discrepancies become more clearly identifiable, pulling esthetic judgments and diagnostic decisions into closer alignment. The study also found that rater characteristics mattered: specialists assigned more critical, centralized ratings, laypeople the most favorable, and significant sex differences appeared in a handful of profiles, with male evaluators rating certain 2D profiles higher and female evaluators rating certain 3D profiles higher.

The authors are careful about the boundary between statistical and clinical significance. With 200 evaluators per arm, even small visual analog scale differences can reach P < 0.05 without reflecting a meaningful perceptual change. They highlight the roughly 15-point gain for the Class III decreased vertical male face as likely clinically meaningful, while the 5-to-6-point gains seen in some other profiles may have limited practical impact, and they note that some non-significant comparisons showed differences of similar magnitude. The power analysis targeted a small-to-moderate effect of 0.29 standard deviations, explicitly a detectability threshold rather than a clinically defined one.

Several limitations constrain how far these findings generalize. The stimuli derived from only two Caucasian subjects, so results may not extend to other ethnic groups or facial morphologies; the rotating animations, while dynamic, do not replicate facial expression, speech, or soft-tissue movement; the online format introduced uncontrolled variation in screen size and viewing conditions; and the occupational composition of the two survey samples differed, with more students and specialists in the 2D survey and more dentists and laypeople in the 3D survey, a confound the authors acknowledge may account for part of the observed modality differences. The sagittal and vertical variations served as standardized experimental conditions rather than variables of direct comparison. Even so, the central message stands: three-dimensional visualization modified perceived facial morphology and was associated with less severe treatment modality selections across a range of patterns, implying that a diagnosis rendered solely from conventional 2D photographs may overestimate skeletal severity and tilt planning toward more aggressive interventions than a fuller, depth-rich view of the same face would support.

Subject of Research: How 2D versus 3D facial image presentation influences perceived facial attractiveness, orthodontic treatment need, and treatment modality selection.

Article Title: Two-dimensional vs three-dimensional facial visualization: effects on perceived attractiveness, treatment need, and modality selection across sagittal and vertical facial patterns

Article References: Chitre Carlisle, M., Botello-Escalante, J., Zhong, K., Kuo, C.-L., Gohel, A., Dolce, C., Yang, X., & Abu Arqub, S. (2026). Two-dimensional vs three-dimensional facial visualization: effects on perceived attractiveness, treatment need, and modality selection across sagittal and vertical facial patterns. Digital and Aligner Orthodontics, 1(1), Article 5. https://doi.org/10.1007/s44525-026-00007-3

Image Credits: AI Generated

DOI: 10.1007/s44525-026-00007-3

Keywords: orthodontics, 3D facial imaging, facial attractiveness, treatment planning, orthognathic surgery, digital dentistry, cross-sectional survey, malocclusion, visual analog scale, clinical decision-making, FaceGen, craniofacial perception