Abstract

The term “eye-limited” resolution (ELR) has seen significant use in recent years within the simulation training and related industries. Results of a literature review revealed several distinct definitions of ELR and a range of estimates of the pixel pitch required to achieve it. When asymptotic visual task performance is used as the basis of ELR, relatively consistent results are obtained for practical tasks such as target identification and orientation detection range. The results of nine published evaluations indicate the pixel pitch that produces 90% of peak performance is in the range of 0.5 to 0.93 arcmin with a median estimate of .7 arcmin. However, a number of authors have asserted that a much finer pixel pitch may be required if observers are to achieve eye-limited performance on hyperacuity tasks such as Vernier acuity. Given that resolution is a primary driver of the performance, cost, and complexity of training display systems, this assertion was tested in the present evaluation.

Performance on the Vernier acuity task was predicted using an observer model and was also measured using five high acuity human subjects who viewed 20 combinations of pixel pitch and antialiasing filter width. The human performance data confirmed the expectation that with sufficient antialiasing a 7.5 arc second Vernier acuity threshold can be obtained with a pixel pitch of 1.6 arcmin. However, a much smaller pixel pitch was required to obtain that level of performance without antialiasing. Based on the results of the research presented here, and our previous work, we conclude that the combination of pixel pitch of approximately 0.7 arcmin and sufficient antialiasing supports eye-limited task performance, even for tasks involving hyperacuity, such as Vernier acuity and stereo acuity. This conclusion is relevant for identifying visual system requirements for training and human factors research.