SceneBlocks is a spec tool, not a page builder. The hard part of an interactive image isn't the code — it's describing exactly where each region sits and what it should do. SceneBlocks makes that description visually, then hands it off.
photo → structured scene model → Lovable-ready package → Lovable builds the page
What the MVP does
- · One uploaded image per scene, stored in your browser.
- · Rectangular hotspots you draw, move, resize and label.
- · One trigger and one effect per region: tooltip, modal, highlight or link.
- · A preview mode where the interactions actually run.
- · An export package: project brief, scene JSON, three phased prompts and a standalone HTML reference.
What it deliberately isn't
- · Not a website builder — one image scene at a time.
- · No accounts, payments, or server storage.
- · No AI segmentation; you draw the regions yourself.
- · No polygon shapes, layers or animation timelines yet.
Coordinates are stored as fractions of the image box, so the exported package stays correct at any resolution or viewport width.