A key characteristic for the effectiveness of Virtual Reality Environments is a high sense of presence -- the feeling of being in that virtual place, even though being physically in another location. For a more differentiated perspective on presence, the term can be defined by subtypes, such as physical, social and self-presence. The Multimodal Presence Scale (MPS) by Makransky and colleagues deals with these dimensions, was translated into German, and evaluated regarding the sensitivity and specificity of the social presence subscale. The results raise the question how well differences in physical and self-presence can be detected by the scale. We conducted an experiment by constructing two virtual worlds, manipulating the polygon count of objects in each world, and measuring presence. Additionally, we assessed the correlation of the MPS with the Igroup Presence Questionnaire (IPQ). No significant differences in physical or self-presence were found. However, when examining an item that closely matches the manipulation of the self-presence world (the user's virtual hands), a statistically significant difference was found. We provide three possible explanations for these results: 1) an insufficient impact of the abstraction levels, e.g., due to insufficient time and attention to the manipulation, or too little difference between the abstraction levels, 2) a lack of sensitivity of the used MPS and IPQ, or 3) the polygon count not being important for physical or self-presence. We conclude that high polygon count might not be that crucial for presence and provide suggestions for future research.