Several studies have focused on the social sharing of visual practices as constitutive of evidence within a domain, while there has been relatively less attention paid to points where the social sharing of practices breaks down, or is resisted. This article argues that a study of both types of cases is necessary in order to gain a better perspective on social sharing of practices, and on what other factors this sharing is dependent upon. The article presents the case of currently emerging inter-disciplinary visual practices in the domain of computational biology, where the sharing of visual practices would be beneficial to the collaborations necessary for the research. Computational biology includes sub-domains where visual practices are coming to be shared across disciplines, and those where this is not occurring, and where the practices of others are resisted. A significant point of difference between these sub-domains is between visualizations that render the output of simulations and those which are images taken during observations using the techniques of microscopy. A crossing over, compromise or sharing of practices relating to these different sub-domains is difficult and often resisted. This resistance needs to be contextualised in a far richer account of the relations between the visual artifacts, the scientists who use them within disciplinary domains, the theoretical and instrumentational outlook of the disciplines in question, and that towards which the science is directed, its domain of study. Social practices alone are not sufficient to account for the shaping of evidence. The philosophy of Merleau-Ponty is introduced as providing an alternative framework for thinking of the complex inter-relations between all of these factors. This philosophy enables us to think of the inter-constitutive relations between these different factors, which ultimately define an epistemological and ontological space in which the object of study itself has an active constitutive role, and in which the scientist as person and perceiving body within a knowledge domain is also constituted.