What question did this study set out to answer?

To develop discrete style transfer as a part-based substitution framework that treats seams and visual discontinuities as structural design features rather than flaws.

August 19, 2026International Journal of Architectural ComputingOpen Access

Discrete style transfer: Syntax, technique, meaning

View Full Paper

Ask AI

Bookmark

Authors

DPDanahy Patrick

Discussion

Member takes

Overview

Computational study demonstrates discrete style transfer via quadtree tiling improves execution speed and stylistic fidelity over continuous models, indicating advantages for modular fabrication.

Key Points

To develop discrete style transfer as a part-based substitution framework that treats seams and visual discontinuities as structural design features rather than flaws.
Subdivided target images into adaptive quadtree tiles.
Encoded spatial tiles into latent embeddings using an autoencoder.
Substituted image components through nearest-neighbor search against style features.
Demonstrated significant improvements in processing speed and stylistic fidelity compared to alternative image comparison methods and mosaic baselines.
Achieved visual similarity competitive with continuous neural style transfer while yielding modular, interpretable outputs suitable for material assembly.

Cite This Study

Danahy Patrick (2026) studied this question.

synapsesocial.com/papers/6a8563d703308d306e2d7362 https://doi.org/10.1177/14780771261478029

View Full Paper

Ask AI

Bookmark

August 19, 2026International Journal of Architectural ComputingOpen Access

Discrete style transfer: Syntax, technique, meaning

View Full Paper

Ask AI

Bookmark

Authors

DPDanahy Patrick

Discussion

Member takes

Overview

Computational study demonstrates discrete style transfer via quadtree tiling improves execution speed and stylistic fidelity over continuous models, indicating advantages for modular fabrication.

Key Points

To develop discrete style transfer as a part-based substitution framework that treats seams and visual discontinuities as structural design features rather than flaws.
Subdivided target images into adaptive quadtree tiles.
Encoded spatial tiles into latent embeddings using an autoencoder.
Substituted image components through nearest-neighbor search against style features.
Demonstrated significant improvements in processing speed and stylistic fidelity compared to alternative image comparison methods and mosaic baselines.
Achieved visual similarity competitive with continuous neural style transfer while yielding modular, interpretable outputs suitable for material assembly.

Cite This Study

Danahy Patrick (2026) studied this question.

synapsesocial.com/papers/6a8563d703308d306e2d7362 https://doi.org/10.1177/14780771261478029

View Full Paper

Ask AI

Bookmark