A self-directed study in coaxing one material language out of Midjourney: ordinary subjects, from a beetle to a goldfish to a human ribcage, rebuilt as if machined from frosted glass, chrome and a warm light trapped inside. The hard part was never a single hero image. It was consistency, holding the exact same glass-and-ember treatment across dozens of unrelated subjects so they read as one designed collection, not a lucky batch.
- 01
Built a reference moodboard of real product photography (frosted resin, medical glass, brushed chrome) and distilled it into a single trained style reference, so every render inherited the same material, lighting and palette.
- 02
Wrote the prompts as a fixed grammar: subject, then *body of transparent cubes / inner neon*, then *chrome plating*, then *clean studio background*, so only the subject changed and the treatment held.
- 03
Ran each concept as a small batch and culled for the cleanest symmetry and the most legible inner glow, the way you would cull a photo shoot rather than accept a first take.
- 04
Animated the strongest specimens and chained them into a morph reel, each object dissolving into the next, to prove the language survives in motion as well as in stills.
A coherent order of around fifty specimens plus a morph film, kept as a style-system proof: evidence that one trained reference and a disciplined prompt grammar can manufacture a consistent design language on demand.
