Industrial SAT formula generation is a critical yet challenging task for heuristic development and the surging learning-based methods in practical SAT applications. Existing SAT generation approaches can hardly simultaneously capture the global structural properties and maintain plausible computational hardness, which can be hazardous for the various downstream engagements. To this end, we first present an in-depth analysis for the limitation of previous learning methods in reproducing the computational hardness of original instances, which may stem from the inherent homogeneity in their adopted split-merge procedure. On top of the observations that industrial formulae exhibit clear community structure and oversplit substructures lead to the difficulty in semantic formation of logical structures, we propose HardSATGEN, which introduces a fine-grained control mechanism to the neural split-merge paradigm for SAT formula generation to better recover the structural and computational properties of the industrial benchmarks. Experimental results including evaluations on private corporate data and hyperparameter tuning over solvers in practical use show the significant superiority of HardSATGEN being the only method to successfully augments formulae maintaining similar computational hardness and capturing the global structural properties simultaneously. Compared to the best previous methods to our best knowledge, the average performance gains achieve 38.5% in structural statistics, 88.4% in computational metrics, and over 140.7% in the effectiveness of guiding solver development tuned by our generated instances.