Æther FacetDesign Bench Start Demo

Design Bench · Step 04 of 04

Iterate

Choose a promising gem, tell the model what you value, and use that preference to guide the next generation of designs. Repeat the cycle until you find the gem you want.

1 · GenerateThe model proposes six candidates
2 · ChooseYou choose the designs you prefer
3 · RefineFuture rounds learn from your choices
Prepared demo, not ML output: All six cuts and the rough-to-facet overlay are illustrative. Geometry is centered and scaled for visualization, NOT registered to an actual scan or verified to fit within the rough. Optical metrics and improvements have not been computed. Saved demo gems are excluded from preference-training datasets.

Your choices will guide the model. Once connected, the learning system will use the designs you select to propose new candidates closer to your preferences. In this preview, your choices can be recorded, but they do not train the model or generate a new batch.

Visual comparison

Rough-to-facet comparison and six candidates

On the left, amber indicates original rough wireframe; the current faceted cut can appear solid or as a cyan wireframe. Selecting a design on the right updates the left comparison.

Current gem · Round brilliantComparison preview
Original rough · amber Faceted gem · cyan
Drag: rotate both · Wheel: zoom · Shift+drag: pan

Loading representative source rough and faceted preview…

Illustrative placement only. The faceted shape has not been fitted to or verified inside this rough.

Candidate 1

Round brilliant

Symmetrical, radial facets · Demo geometry

Candidate 2

Oval brilliant

Elongated brilliant form · Demo geometry

Candidate 3

Emerald step cut

Stepped pavilion geometry · Demo geometry

Candidate 4

Cushion mixed cut

Rounded-square outline · Demo geometry

Candidate 5

Princess style

Square outline, pointed pavilion · Demo geometry

Candidate 6

Marquise brilliant

Lengthwise pointed outline · Demo geometry

Current design · evaluation

Gem metrics

Measured values will appear when the GemML evaluation pipeline is connected.

Awaiting engine
Brilliance—Not measured
Fire—Not measured
Scintillation—Not measured
Contrast—Not measured
Retained volume—Not measured
Entropy—Not measured
Fractal dimension—Not measured

These fields describe the selected design once evaluated. No optical or geometric score is inferred from the illustrative shapes.

Iteration history

Progress across design steps

Compare the selected metric with its starting value over successive optimization rounds.

Metric history

Illustrative example only — not measured. Real results replace this chart when available.

Choose a favorite. Generate the next possibilities.

Select one of the six demo cuts.

Adjust priorities

Sign in and start a session from User Studio to save a gem. Guests can compare demo geometry but cannot save it to the library. Each choice is a preference signal: the model will learn what kinds of designs you select and try to offer better possibilities in later rounds. The demo can record your choices, but generating new ML candidates still requires the engine connection.