Boosting Creativity, Not Just Efficiency

New York is a useful place to have a headquarters for reasons that have little to do with an office. The city offers a constant visual field: façades, entrances, towers, lobbies, scaffolding, old ornament, and new construction. Every so often, a detail separates itself from the background and captures your attention.

Once you begin looking with discernment, certain geometries repeat. Art déco made disciplined use of circles, squares, diamonds, stepped patterns, and angled lines. With enough exposure, ornament starts to feel less like decoration and more like a visual language.

With that observation in mind, Cliffside Studio ran a small design experiment. We wanted to see whether AI could expand creative range across a multi-step production process. The question was less about what AI could make for us than how much more of the design space it could help us explore.

Using only a mobile phone, we took two photographs of an architectural motif at night under less-than-ideal lighting conditions. The images were visually rich but technically noisy, with illumination, reflections, and overlapping decorative elements. They remained working references only, nothing that could be considered part of a finalized product.

The first results were beautiful but elaborate, with metallic surfaces, reflections, and layered decorative systems. They were useful diagnostically, but they were already more than we needed. We wanted to isolate structure rather than polish existing aesthetics.

With simple prompts, we removed texture, framing, and secondary elements until the composition was reduced to black strokes on a white background. At that point, the exercise had moved far enough from the original reference that we could concentrate on geometry itself.

Rhombi — diamonds, if you will — kept surviving those reductions. Eventually, we settled on two sets of three nested rhombi as the minimum viable motif: simple enough to manipulate, but still naturally connected to Cliffside's existing geometric identity.

The Cliffside emblem already had a measured geometric language organized around a recurring diagonal of roughly 31 degrees. Rather than reproduce one of the emblem's existing forms, we used that directional relationship as a constraint.

The selected rhombus was then formalized around acute angles of approximately 58 degrees and corresponding 122-degree obtuse angles. In code, the geometry could be generated from a 29-degree half-angle rather than redrawn manually each time.

The shape was no longer merely something that looked right. It had become a repeatable rule connected to the brand's existing geometry.

The emblem offered another useful cue. Its balance depends on repeated diagonals, controlled mass, and intentional asymmetry rather than perfect symmetry. That principle would later influence the composition on the website.

AI was tasked with exploring motion. The process produced more than 20 animation families, including expansion, contraction, rotation, interlocking, drawing, scanning, compression, perspective changes, and morphing.

The purpose was not to find the most dramatic effect. It was to understand the range of behaviors the geometry could support.

Once that range became visible, choosing among the viable directions became almost secondary. AI widened the field, but widening the field only matters if someone still decides what belongs.

We selected a restrained treatment: two sets of nested rhombi moving in opposing ways.

The instructions were uncomplicated: use the established geometry, remain within the Cliffside palette, construct the form in SVG, and integrate it into the existing landing page.

Scale produced one last conceptual shift. Small, fully visible rhombi behaved like icons. Enlarged and deliberately cropped, they began to read more like architecture. The geometry remained exact, while the composition became asymmetrical and environmental as seen below:

The process followed a method we use often at Cliffside: observe broadly, notice what stands out, identify the pattern, strip away the excess, and build with the distilled material.

What made the experiment interesting was that the resulting object was less an animation than a chain of translations and transformations: a design detail in the city, reverberating across eras, became a photograph; the photograph became a pattern; the pattern became geometry; geometry became motion; and motion became code.