Marwan Mostafa af86b56c77 feat(website): connectable hero canvas + L-to-R layout ...

- Tighten the cluster into a clean left-to-right flow: input on the
  left, canny+depth stacked in the middle column with the slider on the
  green→depth path, final output on the right. Re-tune CONTENT bounds
  for ~80 scene-units of breathing room around the cluster.

- Make the canny→output and depth→output connections interactive.
  Click the CANNY EDGE / DEPTH MAP chip on the final-output card to
  toggle that connection (the chip is exactly where the input slot
  lives, giving a generous click target). Cursor swaps to pointer over
  a chip; disconnected chips dim to warm-gray and their edge line is
  hidden.

- Output mode follows the connection state — both connected shows the
  composed final image (with the existing depth-blur scrubber), only
  canny shows the canny-edge view, only depth shows the depth-map
  view, neither shows an empty placeholder card.

- Keep the canvas wrapper at full bleed and lift just the BrandButton
  to z-20 so drag interactions span the entire hero.

2026-05-07 12:57:39 +03:00

..

desktop-ui

refactor: replace unsafe as Error assertions with type guards (#11845)

2026-05-04 11:40:28 -07:00

website

feat(website): connectable hero canvas + L-to-R layout

2026-05-07 12:57:39 +03:00