Elliot Mairet
- Period
- 2026
- Contribution
- Design and development
- Status
- Live client archive
A photography archive for Elliot Mairet, with an image index, large photograph views, colour-based discovery, and a private publishing workflow. I designed and developed the site; the photographs are Elliot's.
Giving photographs the space
The archive supports three ways of browsing: scanning the collection, viewing a photograph at scale, and following colour relationships. I kept persistent controls to a minimum so each view could give most of its space to the photographs. A dense index lets you scan the collection before choosing an image.
On an individual page, the photograph takes up most of the screen, with its five-colour palette below.
Navigating by colour
Each photograph starts with suggestions based on its overall palette. Selecting a swatch changes the suggestions to photographs with a nearby colour. This gives visitors another route through the archive without a separate search interface.
A five-colour palette is extracted during upload and stored with the photograph. Matching runs in the browser from cached palette data, so changing the selected colour does not need another server request. These are approximate colour relationships: they do not identify the same subject or location. When few close matches exist, the matcher relaxes its threshold to offer up to five candidates.
Adding new work
A private admin area handles uploads, photograph details, and site text. The server validates uploaded image content before publication, with colour analysis built into that workflow.
An image can be accepted even if its publication response never reaches the browser. Restarting the entire upload could then create a second photograph. I separated preparation, file transfer, and publication in the upload controller. After transfer, the controller retains the original paths and metadata so a retry resends the same publication request instead of uploading another object.
The retry regression tests simulate a lost or malformed response and assert that the next request has exactly the original publication payload, even if form values have changed. A separate case checks that a failed transfer restarts preparation rather than publishing an incomplete object. These tests isolate the client controller with request adapters; they do not exercise the storage and database services together.
I used Next.js for the archive and photo routes, with React handling palette selection. The pages can be cached while colour matching runs in the browser. Supabase keeps photograph metadata and admin authentication together; Cloudflare R2 stores the originals. Elliot can manage photographs through the admin, while catalogue updates invalidate cached pages without a site deployment.
The delivered archive connects a broad index, individual photographs, and palette navigation to one publishing workflow. The next question is how useful those approximate colour suggestions are to visitors; the implementation alone does not answer that.