Underwater image-based sampling procedures, using SCUBA, are compared using imagery collected from a temperate hard-substratum community. The effectiveness of a low-budget, high-resolution image mosaicing technique is assessed by comparing the relative efficiency of data collection, extraction and analysis among sampling procedures. In addition, a manipulative experiment tested whether the sampling procedures could detect the physical removal of 10% of the reef community. Overall, four factors were explored within the data: data collection media (stills and video), cover and community composition estimation techniques (visual cover estimation, frequency of occurrence and point extraction), change detection (pre- and post-impact) and depth (8, 14, 18 and 22 m). Stills imagery sampled the reef community at a higher image resolution than the video imagery, which enabled identification of more species and less-conspicuous benthic categories. Using the visual cover estimation technique, the stills imagery also had the greatest benefit in terms of efficiency and species identification. However, the experimental impact was detected using only the point extraction technique. The recommendations are that: (1) the image mosaicing technique is applied to fixed-station monitoring; (2) the point extraction technique be used for efficient and cost-effective monitoring at coarse taxonomic resolutions; and (3) survey depths remain constant over the duration of hard-substratum community monitoring.