Spatial heterogeneity induced by fibrosis strongly influences conduction patterns and arrhythmogenic risk in cardiac electrophysiology. Accurately reproducing the fibrotic microstructure represents a challenge to ensure physiological realism in computational cardiac models. Recent approaches based on Perlin noise are able to generate realistic fibrotic patterns but are difficult to generalize to complex geometries such as atrial tissue. In this work, we propose a mathematically rigorous framework for modeling cardiac fibrosis using Stochastic Partial Differential Equations (SPDEs). This approach allows explicit control over smoothness, spatial range, and oscillation frequency, enabling replication of multi-scale and directional features typical of fibrotic tissue. Crucially, the SPDE framework extends naturally to complex geometries, overcoming the limitations of Perlin noise. We demonstrate that multi-octave Perlin noise can be interpreted as weighted superpositions of SPDE solutions with scale-dependent correlation lengths. Comparative results show strong qualitative and quantitative agreement between Perlin noise–generated patterns and SPDE-based fields, with the resulting fibrotic patterns consistent with clinical observations in spatial organization, anisotropy, and morphological complexity. In conclusion, combining multiple SPDE-driven fields successfully reproduces the diverse fibrotic structures previously generated using Perlin noise. This SPDE-based methodology provides a flexible alternative for synthesizing realistic cardiac fibrosis patterns, improving physiological realism in cardiac electrophysiology simulations and enabling robust modeling in clinically relevant geometries.