This study presents a comparative framework for evaluating emotional engagement with textile soft robots and their augmented-reality (AR) counterparts. Four robotic sculptures were developed, each embodying nature-inspired dynamic behaviors such as breathing and gradual deformation. Using a between-subjects design, two independent groups, one experiencing the physical installations and one engaging with their virtual (AR) twins, follow identical protocols and complete the same self-assessment survey on affective and perceptual responses. This approach minimizes carryover and novelty effects while enabling a direct comparison of sensations such as calmness, curiosity, and discomfort across modalities. The analysis explores how motion, form, and material behavior shape emotional interpretation in physical versus digital contexts, informing the design of hybrid systems that evoke meaningful, emotionally legible interactions between humans, robots, and digital twins.