Architexta was a design-build intensive elective undertaken with students at RMIT in Melbourne. The elective was interested in the intersection of traditional decorative arts and generative AI and explored how existing building surfaces (approximated with cardboard) can be repurposed, resigned and redecorated. Students used Stable Diffusion and Control Net to generate texture maps for physical objects captured through photogrammetry or for digital objects modelled in Rhino. These patterns were then applied manually using mixed reality guides to illustrate where to apply paint or collage intricate clusters of “post its” onto a woven surface. new graphic and architectural languages, and develop techniques for hallucinating systems of ornamentation that ‘fit’ perfectly with the forms on which they are to be applied. The result is a collection of objects that occupy territory somewhere between the qualities of digital generative models and the qualities of objects and ornamentation made by hand.
Project 1 - Applying patterns to scanned objects
Jazz Pedder and Amey Dalvi
Project 2 - Topology agnostic pattern application using post-its
CHEN KE, EBBE VANDENBERGHE, TANAT VIRAPRIYA, TRISHA KARKHANIS, TRISTAN BORROMEO, YAO QINLING, WANG YUSHAN
