Graham Johnson (grahamj.com) discusses challenges in visualizing 'mesoscale' data, i.e. the arrangement of macromolecules within cells - a scale where our knowledge is currently very sparse compared with larger and smaller scales. He presented the uPy and autoFill frameworks he co-developed that can automatically build models on this scale that are consistent with current knowledge, highly informative, and also aesthetically appealing. This talk was presented at VIZBI 2012, an international conference series on visualizing biological data (vizbi.org) funded by EMBO & NIH. This video was filmed and distributed with permission under a creative common license.