 I am Bolin Zhang and I will be presenting work on behalf of me and my co-authors from University of Wisconsin at Madison. As researchers continue to advance methods and systems to produce more accurate haptic renderings, it is also important to consider whether the improved accuracy benefits the perceived realism of rendering. Studying perceived realism can inform the required accuracy and performance of haptic systems. Furthermore, understanding the sensitivities of users to imperfections in model printers can help system designers balance trade-offs during haptic actuator design. We investigated the effect of parameter variations on perceived realism for a common set of kinensetic haptic properties. We identified and modeled a set of four objects consisting of these properties. We used the doll model to simulate the friction in a faucet knob, used the superposition linear stiffness model to simulate the stiffness in a doorknob, and used half and continuous sinusoidal model to simulate the detent in a deadbolt lock and a multimeter knob, respectively. We conducted a user study where participants interacted with a range of model printers for each of the objects and provided corresponding realism readings. One haptic interface provided a variable haptic rendering of the object, where a single parameter of the rendering model was varied from an established nominal value. The other parameter was set at the nominal value and kept unchanged. The other haptic interface provided a nominal rendering of the object to give participants a reference for their realism readings. The parameters of the nominal renderings were collaboratively determined based on a voting procedure among four haptic experts. The nominal renderings were consistent between trials. We fit power model to the realism data, which provides intuition into the sensitivity of the haptic model parameters. We clustered the fit into four groups, which can be principally described by the differences in the power model exponents. The first cluster was FITs, where the decrease in realism was approximately linear. This behavior indicates that participants were equally sensitive to parameter changes across the range of parameter values. The second cluster was FITs, where the decrease in realism was weakly quadratic. The majority of FITs exhibited its behavior. This trend suggests that for these parameters, some inaccuracy in the model parameter may not have a large impact on users perceived to realism. The third cluster was FITs, where the decrease in realism had a higher order exponent in the power model. The higher order exponent indicated that a large range of parameter values near the nominal are perceived as similarly realistic and thus designers may not need to focus on fine tuning these parameters. The final cluster was FITs, where the power model exponent was less than one. Parameters with similar FITs are susceptible to large decreases in perceived realism for small parameter changes and thus may require fine tuning. The results suggested that changes to the perceived realism are dependent on the haptic property that is varied. Future work will focus on examining a wider range of objects and haptic rendering effects. In addition, we plan to investigate parameter interaction to further understand human sensitivity to rendering parameters.