INGRID starts to see: a camera pipeline and Vision panel

The first real step toward INGRID perceiving more than telemetry and text: a camera input pipeline and a dedicated Vision panel in BlueCrow Studio.

• Camera input pipeline. A small, provider-neutral ICameraCaptureService abstraction (mirroring how INGRID’s existing hardware-telemetry and music-generation providers are shaped), backed by a real Windows implementation built on FlashCap — a pure-managed, actively-maintained capture library chosen specifically because it needs no MSIX packaging, unlike the WinRT camera APIs, and BlueCrow Studio ships as a plain unpackaged desktop app. Device enumeration, live frame capture, and disconnect handling are all real — a capture failure or an unplugged device turns into an honest Error/Offline status, never an unhandled exception.
• INGRID Vision panel. A new panel showing the live preview (or an honest “no camera detected” / “connecting” / “error” state), camera status, real measured frame resolution and FPS, and clearly-labeled “coming soon” placeholders for object recognition, scene understanding, facial tracking, text recognition, and emotion recognition — none of which are implemented yet, and none of which are ever implied to work today.
• A new Observing avatar state. INGRID’s avatar gains a fifth state alongside Idle, Listening, Thinking, and Speaking — a distinct visual treatment for when she’s actively attending to the camera feed, surfaced automatically whenever the Vision panel is open.

No physical camera exists on the machine this was built on yet — enumeration honestly reports zero devices, which is the correct, expected, tested result today. The pipeline is validated with a fake injectable camera source (proving the preview, FPS measurement, and disconnect-handling logic genuinely work) and with the real enumeration path (proving it degrades honestly, without fabricating a detected device). The moment a real camera is connected, no further code change is needed for it to show up.

Not built yet: any actual AI vision capability, and the richer eye/head-movement avatar rig envisioned for a more animated INGRID presence — tonight’s avatar work stays within the existing rendering system, not a new one.