Dibei gas field is located in the northern structural belt of the Kuqa Depression, hosts the Ahe Formation as its primary reservoir. The Ahe Formation is characterized by a sand-to-mud ratio exceeding 80%, classifying it as a typical high sand-to-mud ratio reservoir. Dominated by braided channel deposits, the reservoir features thick sand bodies with multi-stage stacking and erosional amalgamation. Conventional modeling techniques struggle to accurately delineate the geometric morphology and stacking patterns of individual sand bodies, leading to inadequate representation of reservoir heterogeneity. The Kuqa River outcrop profile was leveraged to construct two types of training images: object-based and 2D digitized cross-section-based approaches. Correspondingly, the single normal equations simulation (SNESIM) method and the sequential bi-dimensional conditioning data + direct sampling (s2Dcd+DS) method were employed to develop 3D sand body models for the Dibei gas field. The modeling results were evaluated through visual analysis, quantitative variogram parameters, and well data sparseness validation. The results demonstrate that under high sand-to-mud ratio conditions, the s2Dcd+DS method (based on 2D digitized cross-sections) produces 3D sand body models that better align with the actual geological characteristics of the study area. This method effectively captures the geometric configurations and stacking relationships of sand bodies in high sand-to-mud ratio settings, providing a robust foundation for the precise characterization of reservoir heterogeneity. The proposed workflow offers a novel approach for 3D sand body modeling and heterogeneity characterization in high sand-to-mud ratio reservoirs, advancing the integration of geological realism into reservoir models.