Exploring the limits of the fidelity of visual transcriptions.