01 / Government
Public responsibility
Heritage continuity, public space, environmental responsibility, and a legible relationship with the city.
AI Green Nexus City
懸浮綠境 · AI未來城
An urban design study combining simulated government, developer, and resident perspectives in scheme selection, with AI-assisted development of architecture and landscape.
Design process: simulated negotiation, scheme comparison, and AI-assisted development.
The workflow combines the brief, environmental information, simulated negotiation, and computational analysis. The two diagrams trace the steps from setting objectives and generating alternatives to comparison and design development.
Evidence, objectives, and simulated decisions enter the same workflow.
Generate alternatives, then compare and refine them for further development.
Simulated government, developer, and resident perspectives.
At Wuyi Road 155, I organized design decisions through simulated negotiation. Government, developer, and resident perspectives brought priorities of public responsibility, operational value, and everyday use, alongside shared questions of historic continuity and environmental quality.
After selecting a scheme, I used AI to explore variations in architectural form, facade, and landscape, checking the generated images against the spatial design.
Comparing needs through simulated negotiation.
02 / Developer
Operational value
Program mix, usable floor area, vertical farming, and a viable relationship between investment and use.
03 / Residents
Everyday livability
Access, views, open space, human-scale movement, and the feeling of living with the new intervention.
Massing comparison across competing needs.
Simulated government, developer, and resident perspectives informed scheme comparison, addressing building volume, preservation, visual corridors, and the allocation of public and productive space.
I used EvoMass to generate candidate massing schemes, then compared them against these needs, selecting and recombining spatial features.
Stage the voices
Set up the government, developer, and resident positions as a shared decision frame.
Generate
Use parameters and EvoMass addition or subtraction to generate candidate massing schemes.
Compare
Read program fit, sight corridors, public access, and environmental conditions together.
Select
Retain spatial features suited to the site’s needs and recombine them for further development.
EvoMass opens the search with a fast genetic pass.
The Grasshopper system keeps the search adjustable.
Different parameter weights generate five massing alternatives for comparison.
Select a scheme for further development.
After selecting the massing, use AI to explore facade and landscape alternatives.
AI-assisted facade and landscape exploration.
Using the selected massing as a basis, I explored facade, green-corridor, and landscape alternatives with AI.
Generated images supported design discussion and visual studies; they cannot substitute for construction documents or performance validation. I checked their consistency with the selected massing, program, and spatial relationships before retaining useful design ideas.
Start from the massing
Use the selected massing and its spatial relationships as generation inputs.
Translate
Explore architectural form, facade language, planting, material, and atmosphere.
Review
Check whether each image still supports the chosen program, movement, and urban relation.
Advance
Bring useful visual ideas back into the proposal, drawings, and final communication.
Comparing architectural treatments through AI-generated images.
Exploring facade language from the selected massing.
Facade design and its relationship to the street.
Program, section, and surrounding blocks.
The final proposal combines needs identified through simulated negotiation with design ideas explored using AI. The drawings show program, circulation, vertical organization, and connections to the surrounding blocks.
Functions, access, and vertical organization.
Openings and movement remain legible in the selected massing.
Planting, business, office, and parking share one section.
The proposal returns to the surrounding block structure.
Architectural views and spatial details.
Three close views show the building edge, street frontage, and interior spaces.
Green infrastructure becomes part of the building edge.
The proposal negotiates a new edge with the existing street.
Productive and public space share a transformed interior.
The project moves from negotiation to a spatial proposal.
Final presentation board.
The complete board brings together simulated decisions, computational studies, AI-assisted development, and the final scheme. A single-page PDF is available below.
