Abstract: Talus deposit is often seen in hydropower projects in the southwest of China. Its mechanical properties are so complex that the microstructural study is often used to reveal the essence of its deformation and damage. The digital image processing technologyDIPT) is introduced to the simulation of talus deposits. Based on the study of DIPT, a Photo-To-FLAC^3DPTF) auto-modeling program is developed. It is able to realize the whole process: analyzing and processing the digital images, acquiring the information and establishing the micro- model files. A new modeling method is developed for the mechanical behaviors of talus deposits. As an example, some talus deposit data of Gushui Hydropower Station are used. A micro-model for the talus deposit is established fast and correctly by means of the PTF from the field digital photo. The mechanical properties and parameters are studied by the numerical simulation of triaxial tests. The results show that the talus deposit has the feature of “bully” in deformation while that of “Unicom band” under force. A random statistical analysis method for the mechanical parameters is proposed by selecting field multiple photos. The mechanical parameters based on the statistical analysis method are very close to those of experiments in the actual engineering, indicating that the new method of acquiring mechanical parameters is credible. A new way is opened up for selecting the mechanical parameters of talus deposits, which pr ovides a useful supplement for large-scale field tests.