Abstract: In the low-carbon transformation of the new type of power system, providing high-quality, targeted new energy scenario data plays a crucial role in the planning and construction of the system. To address the issues of feature entanglement and limited generalization ability in traditional wind and solar scenario generation methods, this study proposes a controllable generation method for wind and solar power output scenarios based on Conditional Style Generative Adversarial Network (CStyleGAN). Firstly, this study introduces a mapping network into the conditional generation network to decouple the different environmental conditions of the original new energy power stations and generate multi-level style control parameters. Subsequently, a progressive generation structure is adopted in the network generator, thereby refining the features by processing the style control parameters in a hierarchical order through multiple upsampling operations. Finally, based on actual data from Washington State, using the proposed method, actual wind and photovoltaic power station data are simulated and generated. Compared with two benchmark models, it achieves superior performance on both temporal and spatial feature evaluation indicators and can generate targeted new-energy scenario data according to specified environmental condition information.