The historical but questionable size-distance invariance hypothesis (SDIH) features computation over geometric, oculomotor, and binocular cues and the coupling of percepts—perceived size, S′, is mediated by perceived distance, D′. A contemporary non-mediational hypothesis holds that S′ and D′ are specific to distinct optical variables. We report two experiments with an optical tunnel, an arrangement of alternating black and white concentric rings, that allows systematic manipulation of the optic array at a point of observation while controlling a variety of size and depth cues. Participants viewed targets of different sizes at different distances monocularly, reporting S′ and D′ via magnitude production. In Experiment 1, the target was either placed in a continuous tunnel (extending 164 cm) or in a tunnel that truncated at the target's location. Experiment 2 included a third tunnel, one that was truncated with a flat depiction of the posterior surface structure that would have been visible in the continuous tunnel. In both experiments, S′ decreased with D but D′ was unaffected by S. Partial correlation analyses showed that the relationship between S′ and D′ was not significant when the contributions of other variables were removed. Importantly, S′ and D′ were affected differently by manipulations of the optical tunnel's continuity while computationally obvious visual cues were controlled. These outcomes suggest that D′ is not a mediator of S′. Rather S′ and D′ are independently determined with correlated but different optical bases, results that support the direct model.