参数化与智能化协同的景观生成式设计OA

[Objective]To address the insufficient spatial cognition and weak design-logic construction in existing generative models for landscape architecture,this study aims to develop a collaborative generative design method integrating parametric and intelligent approaches.[Method]Taking the Rockery Garden of the Health and Wellness Center in Jincheng City as a case study,parametric layout generation of rockery elements was first implemented on the Grasshopper platform using algorithms such as Random and Voronoi,forming an interpretable and controllable spatial structure.Subsequently,with the support of the ComfyUI platform and the Stable Diffusion model,rapid generation of multiple design schemes and diversified visual expressions were achieved through LoRA fine-tuning and the ControlNet module.[Result]The parametric method efficiently generated multiple layout schemes while maintaining stable spatial structural logic.The Segmentation module effectively enhanced planting configuration and spatial layering while preserving the original rockery layout.When the CFG scale was controlled within the range of 8-10,the diffusion model achieved a favorable balance between stylistic expression and image quality.[Conclusion]The proposed generative design framework characterized by rule-based controllability and diverse generation enables a continuous design process from spatial structure generation to landscape style expression.It provides a technical pathway for generative design in landscape architecture that is interpretable,controllable,and efficient.