Data-driven methods for predicting and evaluating microstructural characteristics have attracted significant attention owing to recent developments in artificial intelligence (AI). In high-temperature deformation processes, the microstructure changes nonlinearly, according to the causal relationship with the deformation history in the form of time series.This makes it difficult to predict the microstructural evolution using existing AI generative models, in which time-series data are not used as an input. Therefore, herein we proposed a novel method to establish a connection between the time series deformation history and the latent vector of the AI generative model. To this end, the dynamic recrystallization (DRX) fraction and DRX grain size of a microstructure were calculated based on the deformation history, which included the temperature, strain, and strain rate, using the finite element method (FEM) combined with a DRX kinetic model. By applying the calculated DRX fraction and DRX grain size as label data, a conditional deep convolutional generative adversarial network was trained to generate microstructures. It was confirmed that the microstructural evolution due to the deformation history can be realistically reproduced. Furthermore, by comparing the average grain size and grain size distribution of the synthetic and actual microstructures, it was proven that the proposed model can be used to accurately predicts not only the shapes but also the quantitative features of microstructures. The results of this study demonstrate that FEM and AI technologies can be used sequentially to simulate the microstructural evolution as photorealistic images.