Parametrično generiranje standardiziranih prostorov

The architectural design process for residential developments faces a critical inefficiency where traditional workflows require weeks for preliminary designs and months for detailed development, creating barriers for smaller investors and overwhelming architects with repetitive technical tasks. This thesis presents the Parametric Generation of Standardized Spaces, a comprehensive system that bridges the gap between user intent and Building Information Modelling (BIM) through the integration of natural language processing, building code compliance, and direct BIM generation. The system architecture comprises three interconnected components: an external user interface with custom natural language interpretation, a PyRevit plugin implementing parametric generation within Autodesk Revit, and a robust file-based communication protocol. The implementation successfully incorporates verified building code requirements from ten countries, enabling proactive compliance during generation rather than post-generation verification. Through template-based parametric adjustment and sophisticated area distribution algorithms, the system generates fully detailed, code-compliant apartment BIM models within seconds, replacing processes that traditionally required weeks. Testing demonstrated successful generation of diverse configurations from studio to four-bedroom layouts, while maintaining architectural quality and regulatory compliance. By democratizing access to professional-grade design tools, the system empowers investors, clients, and architects alike to explore and generate architectural solutions without technical barriers. This research validates that natural language interfaces can successfully translate human spatial thinking into complex BIM operations without sacrificing professional standards, fundamentally transforming how standardized residential spaces are conceived, designed, and delivered.