The options
Current — photograph (approved 25 September 2026)
The supplied 1880 × 836 photograph, banner placement, static. Warmest and most human of the three, and the only one that shows people.
Open questions it carries: the source file name shows it was generated with ChatGPT, and the dashboard inside the image reads 92% + 12% + 3%. Both are recorded against DR-18 and the hero artwork row of the R2 README.
Open the approved homepage
Option A — Drivetrain
The gear cluster from the approved kit: a large silver wheel, an orange pinion and a small idler, joined by hairline shafts with teal drive joints, plus the corner crop. Each gear is geared to the same scroll and idles at a speed inverse to its size, so the mesh stays believable when the page is still.
Added for this option: an engage sequence on load (the gears settle, the shafts draw from the gear they carry, the joints arrive), a hairline ring on each joint so it reads as a joint rather than a dot, and one pass of light round the wheel rim as the mechanism takes load.
Why it might win: it is the brand's own idea — machinery of transformation — rather than a stock scene, it needs no photography licence or approval, and it is the only option that says something about how the firm works.
What it costs: the gear PNGs are 476 and 525px and render up to 672px, so they soften on 2× screens until the client supplies larger marks (DR-02). The cluster is hidden below 1024px, so phones get the copy alone.
Open option A
Option B — Instrument
The code-built interface panel already used for the flagship product, promoted into the hero: a raised white sheet carrying a governance ring, a financial sparkline, a risk heat strip and a portfolio timeline, with one quiet gear behind it as the brand anchor.
Added for this option: the approved reveal choreography (bars grow, the sparkline draws, the ring sweeps), then a live layer — a slow scan pass across the sheet, a drifting "today" marker, and a gentle lift as the hero scrolls, where the browser supports scroll-driven animation.
Why it might win: it shows the product rather than a stock idea of teamwork, it is drawn in CSS so there is no photograph to licence, no AI-generated people, and every label is a category word or deliberately blank — there is no figure in it that can be wrong.
What it costs: it is abstract. A buyer who wants to see faces will not find them here, and the panel must never be presented as a screenshot; the caption stays data-pe-sample.
Open option B