Abstract: To improve the performance of loading city level oblique photography models with massive data scale on web pages, this paper used techniques such as dynamic link library conversion, multi-page model merging, and multi-threaded model compression to lightweight process the source data of oblique photography models, optimized the query logic and loading queue during the model loading phase through methods such as scene roaming program logic optimization and model data loading process optimization, achieving smooth loading of massive models. The research results have verified the feasibility of massive model web loading technology, which has reference significance for the improvement of related technologies in the railway industry.