ELIZIUM / Audience-response platform / 2026
ELIZIUM
A website for an immersive experience that continues beyond the encounter. Authored questions invite a response; the archive makes those reactions visible as a collective pattern.
Working MVP developed for a confidential company. Generated imagery. Physical experience proposed.
What I developed
I developed the concept, question archive, visual language and public response flow. I specified how reactions should be recorded, aggregated and reviewed, including the operator view, enquiries, connection states and insight tools. I directed the assisted Next.js build, reviewed the implementation and drove revisions through interaction and integration tests.
Next.js · React · TypeScript · Tailwind CSS · Framer Motion · Claude-assisted development · Vercel
From a question to a collective response
Each signal carries an authored question, an image and a date. Visitors choose anxiety, interest, trust, discomfort or emptiness. The response flow records the choice, then shows an aggregate summary rather than a claim about an individual’s emotions.
The recorded test
Seven response events were recorded in the original test: interest 4/7 (57.14%), anxiety 1/7, trust 1/7 and discomfort 1/7 (14.29% each). These are test events, not verified unique visitors or current audience analytics.
The website has an operating layer
The response and enquiry routes connect the public interface to Make and Google Sheets, with Gmail for communication. I specified GA4 and Microsoft Clarity for behaviour review, and Notion for organising project information. The operator view distinguishes Live and Fallback using the API’s returned mode, and exposes missing archive dates and pending reports.
I tested the path from a reaction to its summary and reviewed connection failures. A failed connection must remain visible rather than presenting a local fallback as a live result. External integrations have not been freshly validated, so these original records demonstrate the build and testing, not an assurance of current service availability.
Make · Google Sheets · Gmail · Google Analytics 4 · Microsoft Clarity · Notion · API integration and recorded test review
Related film research
Alongside the platform, I developed the narrative, characters, visual canon and continuity criteria for a generative film. I directed ComfyUI workflows and local repair decisions, including rejecting a depth-conditioned result that preserved motion but lost costume detail. This is separate, ongoing film development.
ComfyUI · RunComfy · Wan / VACE · DepthAnything V2 · SteadyDancer · Krea · Higgsfield · Assisted local repair