Cameras back in a smaller form?

[Mika2108]

Hi! Since NextFloor is in daily use on our wall tablet, the one thing we really missed is the cameras, so we brought a smaller version back in our own copy. Before opening anything, I'd like to ask: you removed cameras on purpose in 0.8.0 – would you take them back in a reduced form, or would you rather keep them out of NextFloor?

What we have, built on the code you removed (c605f8e) and kept small:

• Camera as a device kind: wall or dome model, a field-of-view wedge on the floor that stops at the walls, red while a motion / occupancy / presence sensor of the same device detects something.
• Live window in the bottom-right corner: a tap on the camera or its wedge opens Home Assistant's own player (picture-entity with camera_view: live, so WebRTC/HLS through go2rtc, MJPEG as a fallback). ✕ or a second tap closes it, ⤢ opens HA's big dialog. Card option camera_popup: corner | dialog.
• Wedge settings in the camera's form: brightness, whether a tap on the wedge opens the camera or goes through to the room, which detections turn it red (person, vehicle, animal, motion), and the tilt: with the height, it decides where the view meets the floor.
• Alarm panel as a device kind: shield marker with HA's state text, red when armed or triggered, orange while arming or in the entry delay; a tap opens HA's code keypad.

Not brought back: the camera wall, flying into the picture, the motion trail and the detection pins.

If you're open to it, I'd send it as one PR with small separate commits (camera kind and live window, alarm, wedge settings, tilt, detections), each with tests, and the texts in German, English and French. If you'd rather not have cameras in the core, would a plugin hook for taps be an option, so this could live as a plugin? Either answer is fine with us.

[therealMRBK]

Thanks for asking first, and for keeping it small!

I'd rather keep cameras out of the core and have them as a plugin. A plugin hook for taps is the way I'd like to go: if you send a PR with the hooks you need (a tap handler for a device kind, plus whatever the live window and the wedge need from the view), kept generic and with tests, I'm happy to take it. Your camera and alarm plugin can then live in its own repository, and I'll link it from the README.

[therealMRBK]

Small change of plan: I went ahead and built it myself, along the lines you described, so you don't need to send anything.

• Plugin API (on dev, ships with 0.15.0): device extensions can now add fields to the device's form (settings), draw a wedge on the floor that stops at walls and can be tapped (floorWedge), take a tap and open a small window in the corner of the 3D view (tap with showWindow / moreInfo), and name further entities to watch (watch). Devices a plugin takes on are drawn in the editor too, and a device with a wedge gets the aim handle. Docs: docs/plugins.md.
• Cameras plugin in the plugin store: the field-of-view wedge (red while a motion, person, vehicle or animal sensor of the same device detects something; which ones count is a setting), the live picture in a corner window through HA's own picture-entity player (WebRTC/HLS via go2rtc, MJPEG as a fallback) or HA's dialog, mount, field of view, reach and tilt in the camera's form, and alarm panels with their state and the keypad on a tap.

If you try it on your tablet, I'd love to hear how it does, and if you miss something from your version, a PR against the plugin is very welcome.