Multimedia capture

The capture flow is the heart of Lumos Capture in the field. A single sequence groups photos, an audio note, and optionally a text note into a coherent capture that the AI will transform into a draft narrative. This page describes each step.

Start a capture

On an inspection’s detail, tap the floating + button orange at the bottom right. The capture flow opens directly on the camera.

If it is your first capture, iOS asks for camera access permission. Tap Allow (see Lumos Capture — install and get started).

Step 1 — Photo capture (multi-shot)

The full-screen camera opens.

The interface

Counter at the top center (“0/6 photos”, maximum 12 photos per capture).
X button at the top left to cancel.
Flash button at the top right (tap to switch on / crossed-out lightning for off).
Right-side buttons: flip camera, grid mode, zoom + and zoom −.
Zoom selector at the bottom: 0.5x (wide) or 1x (standard).
Capture button round white in the center — tap to take a photo.
Next button orange on the right (with arrow →) — to move to the next step.

Take multiple photos in a row

On each tap of the capture button, a thumbnail appears at the bottom of the screen. You see in real time the photos already taken, with a small red X button on each to remove if needed.

You can take up to 12 photos in a single capture. This is useful to document a component from multiple angles, capture wide and close contexts, or include identification details (serial number, nameplate).

When to tap Next

Once all your photos are taken, tap Next to move to the note screen. You can also continue without photos by tapping Next immediately, but in practice a capture with no photo is not the norm.

Step 2 — Add a note (audio or text)

The Add a note screen appears with:

At the top: photos recap (stacked thumbnails + “N photos”).
Audio note / Text note toggle — you choose the mode.
Skip button (gray) at the bottom left to finalize without a note.
Done button (orange) at the bottom right to finalize with the note.

Audio note

Default mode. Very useful in the field — you describe what you see out loud in seconds.

Record

Large orange microphone button in the center.
Text “Tap to record” below.
Tap the microphone button to start. A timer appears.
Tap again to stop.
If it is your first use, iOS asks for microphone access. Tap Allow.

After recording

Green checkmark in the center.
Text “Recording saved (00:13)” with duration.
Re-record button to start over if you are not satisfied.

Text note

If you prefer to type:

Tap the Text note toggle.
“Add a note…” text field with iOS keyboard.
Type your observation freely.

To use when:

The environment is noisy (nearby construction, strong wind).
You are in the presence of third parties who do not consent to audio recording (see Privacy — Audio).
You want to choose words precisely for technical findings.

Finalize

Tap Done (orange) — the capture is saved and sent to the AI pipeline.
Or Skip — the capture is saved without a note, just the photos.

Step 3 — AI processing in the background

You automatically return to the inspection detail. Several things happen simultaneously:

The “Processing” counter

The Processing card increments to 1 (or increments if you capture multiple findings successively). An orange loading icon spins next to it.

If you are offline

The counter shows “Stored, waiting for network” below the number. Your photos and notes are kept intact locally. As soon as connection returns:

Lumos Capture automatically sends the capture to the AI pipeline.
The counter updates to reflect the sync.
AI processing starts.

You have nothing to do — sync is invisible.

The AI pipeline

The AI takes your capture (photos + transcribed audio + text note) and generates a draft narrative:

Title (e.g. “Leaking kitchen faucet”).
Structured description (Identification, Method, Consequences, Recommendation per the template).
Initial classification: Section, Sub-section, Type (Information / Limitation / Deficiency / Method), Severity.

This processing typically takes a few seconds to about twenty seconds, depending on the length of your audio note and the connection. You do not need to wait — you can continue to capture other observations while the AI works in parallel.

Once processed

The narrative appears as a card in the inspection’s narrative list:

AI-generated title.
Photo counter + audio icon if voice note.
Timestamp (“1m ago”).

The To review card increments. Tap the narrative to open the Review modal and adjust the content (see Review findings).

Best practices

Photograph the wide area first, then the detail — to situate the finding in its context before close focus.
Describe in natural language — the AI handles everyday language well. No need for a script. Describe what you see, where it is, what catches your attention.
Mention section and sub-section in your voice note — for example “Plumbing, kitchen sink, leak at the faucet” — to help the AI classify correctly.
Capture in coherent small batches — one finding = one capture. If you observe two distinct issues in the same room, do two separate captures.
Check the Processing counter before leaving the site — to make sure no capture is still queued without connection.