Thumbnails:
List:
Year:
Category:
Session:
Poster:
Getting poster data...
Philip Hubbard (Howard Hughes Medical Institute, Janelia Research Campus, Ashburn, VA, USA)
High-quality videos increase the impact of scientific research in technical publications and popular media, yet making such videos is difficult. The neuVid system simplifies the production of videos from neuroscience and cell biology data by taking high-level descriptions as input. Descriptions are structured versions of statements like: “load two sets of neuron meshes from a server,” “frame the camera on the first set,” “orbit the camera for 10 seconds,” “5 seconds in, fade out the second set.” These commands are converted to operations in Blender (for segmentations) or VVD Viewer (for direct volume rendering), which then produces video with smooth animation and high-quality rendering. Basic versions of descriptions can be imported from Neuroglancer (and other interactive web applications that use it, like neuPrint, FlyWire Codex or OpenOrganelle), and complete descriptions can be generated from natural language using a neuVid desktop application powered by a large language model (e.g., OpenAI GPT-4).