Direct numerical simulations of fluid flow on three-dimensional pore-scale microstructures derived from images promise more, cheaper, and faster, special core analysis for estimating volumetric and transport properties of rocks. However, the micron-scale X-ray computer tomography images generated by the present imaging technology are limited in resolution, and thus, a significant portion of rock pore volume can remain unresolved. The missing pore volume is not accessible to direct numerical simulations which limits applicability of digital rock physics to infer true residual saturation of reservoir fluids. To derive meaningful results from direct simulations, at minimum, raw fluid saturation inferred from simulations on micron-scale images must be corrected for the missing pore volume. We use concepts of capillary physics in rocks to quantify the impact of image resolution on image-derived fluid saturation and develop novel transforms that compensate for this effect on estimates of fluid saturation from multiphase simulations without the need for higher-resolution imaging. We find that image resolution constraints provide quality indicators when comparing digital rock-derived fluid saturations (e.g., connate water saturation) with those measured in a laboratory.