Social media messaging has long been harnessed to inform faculty about their respective learners. The textual channel is often used because of the ease of interpretation and analysis. Social imagery—tagged images, #selfies, grouped imagery, and others—has been less used, in part because images are more complex and multi-meaninged to analyze. Also, there are not many generalist models that inform how to code or even understand social imagery in an emergent way. (There are large-scale computational means to interpret online images, such as the AlchemyAPI of IBM Watson, for various types of feature extractions. There are ways to code imagery based on specific research questions in particular fields-of-practice.) The presenter recently analyzed a 941-image #selfie + #humor image set from Instagram, with three main research questions: What does identity-based humor look like in terms of a #selfie #humor- tagged image set from the Instagram photo-sharing mobile app? Do more modern forms of mediated social humor link to more traditional forms theoretically? Is it possible to apply the Humor Styles Model to the images from the #selfie #humor Instagram image set to better understand #selfie #humor? What are some constructive and systematized ways to analyze social image sets manually (with some computational support)? This digital poster session will highlight some of the initial research findings (forthcoming in a near-future publication) and share insights about effectively coding social imagery in a bottom-up and emergent way.