The effect of penetrating and non-penetrating fluids, and of opening shape, on hydraulic fracturing has been extensively studied, both theoretically and experimentally. These factors are important because of their effect on fracture initiation and breakdown pressure. This paper revisits these problems, but with a novel experimental procedure that allows one to see details of fluid penetration and fracturing. Prismatic gypsum specimens were tested under external biaxial loads and with hydraulically pressurized openings. Unlike previous experimental research, that used different materials, fluid viscosities and/or pressurization rates to vary the fluid penetration, this study keeps these testing conditions constant and only modifies the permeability of the material around the opening by sealing or not sealing its interior faces with wax. Most importantly, while previously fluid penetration and fracture initiation and propagation patterns could only be visually observed after completing the experiments, the imaging techniques employed in this paper enable us to see the entire fracturing process in detail through high-speed video. Four types of geometries were tested, namely a 5 mm circular flaw, a single vertical ovaloid flaw, and 5 mm circular openings with short and long notches (2.5 mm and 5 mm length, respectively). The first two shapes were analyzed with both, penetrating and non-penetrating fluids. The results confirm that fluid penetration lowers breakdown pressure, does not affect the fracturing pattern, and influences the difference between fracture initiation and breakdown. Also, the high-speed imaging makes it possible to see a distinct fracturing behavior (stable, unstable) within the sealed and not sealed zones. The main contribution is that these effects are not only inferred but are clearly observed during the tests.