Abstract: One of the largest goals of interactive computer graphics is to allow a user to freely walk around a virtual recreation of a real environment that looks as real as the world around us. But hand modeling such a virtual environment is inherently limited and acquiring the scene model using devices also presents challenges. Interactively rendering such a detailed model is in general beyond the limits of current graphics hardware, but image-based approaches can significantly improve the status quo.

We present an end-to-end system for acquiring highly detailed scans of large real world spaces, typically consisting of forty to eighty million range and color samples, using a digital camera and laser rangefinder. We explain successful techniques to represent these large datasets as image-based models and present contributions to image-based rendering that allow these models to be rendered in real time on existing graphics hardware without sacrificing the high resolution at which the datasets were acquired.