<![CDATA[This study develops a field-to-ornament framework connecting agricultural field geometry, computational jewelry design, and evidence-grounded interface storytelling. The main design pipeline uses 350 Fields2Benchmark polygons from the Netherlands, Lithuania, and Estonia as authoritative sources for contour translation. Five abstraction levels are evaluated through geometric similarities and a Geometry Preservation Score (GPS). The Balanced setting retains mean GPS 93.46, compared with 95.06 for Subtle and 88.06 for Abstract. The non-Trace ranking persists across 15,000 sampled weight–scale combinations, although orientation-only weighting reverses Subtle and Balanced. Balanced remains a heuristic demonstration choice without evidence of aesthetic preference. Optional image-recovery experiments assess the limitations of deriving contours when vector geometry is unavailable. On 53 overlay-suppressed annotated map plates, Random Forest achieves mean IoU 0.5417, boundary F1 0.2119, and absolute area error 53.56%. A separate 350-tile AI4Boundaries subset yields modest labelled-extent recovery on 70 aerial test tiles; contour smoothing does not correct prediction errors. Four provenance-card designs are instantiated for 30 fields, producing 120 deterministic audits. Balanced disclosure scores 20.00 and the unsupported minimal claim 100.00 under the proposed overclaim rubric. Sensitivity and controlled content–layout comparisons characterize the rubric's behavior, without measuring comprehension or trust. The framework documents how field geometry can inform two-dimensional ornament concepts while distinguishing design provenance from claims about crop identity, ownership, farming practice, material origin, or environmental benefit.]]>
Copyrights © 2025
