2.1 KiB
Nikos Salingaros' initial prompt:
I need to develop an architectural form language for Ecuador. The exercise needs to locate open-source images of traditional buildings and urban spaces and synthesize those into a design grammar. This grammar could then be recombined within the constraints of the brief to produce a new design. Consider the climate of Ecuador, which is temperate all year around with a fixed amount of rainfall. The form language should choose from the entire spectrum of buildings dating back to indigenous wooden sheds with conical roofs, to stone pre-Columbian pyramids, to early Spanish colonial wooden structures, to highly baroque later Spanish colonial buildings that adapted to the local climate and materials. Pay special attention to the large variety of colors used for centuries in Ecuadorian architecture. Also recover the different ornamental traditions in Ecuadorian and neighboring Latin American architecture. What can you propose for this project?
Nikos Salingaros' follow-up prompt:
You have the right idea. Specialize on the Quito climate and forget the rest, since this is where the project is to be built. But draw from all other regions for the form language. For the copyrighted images, we are not going to copy them and publish them, but will only use them as references for the developed Ecuadorian form language. What I need from you is a detailed document to use together with other instructions in a future prompt to an LLM to produce images of the proposed new building. Other elements of the prompt are going to include the brief constraints, spatial chassis, and other specifics. The climate is fixed at Quito sierra. Color and ornament are not going to be specified, since we will ask for a variety of projects to be generated, each one with a distinct choice for color and ornament to cover traditional Ecuadorian architecture. Then a human will choose the most beautiful from among those. Since this form language document will be used as part of a prompt, the structured grammar document is more useful than generated images. Those will be generated after the LLM creates the verbal descriptions from the prompt constraints.