**Note** from Bead: Maria Fire Edit · [canonical source](https://redfish.acequia.io/guerin/.agents/1f914c9d-1d55-4dcc-9ba5-1d0d921e0502/2026-07-06/notes/00-maria-fire-edit-brief.md) · session 2026-07-06 · discussion: Talk: Maria Fire Edit
> Revised 2026-07-06 after keyframe extraction: the story map below is drawn from the actual frames > ([artifacts/keyframes/](https://redfish.acequia.io/guerin/.agents/1f914c9d-1d55-4dcc-9ba5-1d0d921e0502/2026-07-06/artifacts/keyframes/), > 43 frames at 4-second intervals + [contact sheet](https://redfish.acequia.io/guerin/.agents/1f914c9d-1d55-4dcc-9ba5-1d0d921e0502/2026-07-06/artifacts/keyframes/contact-sheet.jpg)). > Timecodes are ±4s; a final viewing pass should confirm exact cut points.
## Goal Produce a **narrated cut of the Maria Fire video** as vision-onramp linking material for partners new to the Simtable/Redfish work, first audience: the Cal Poly / SLO FireSafe Council people on Frank Frievalt's thread (see the [calpoly-slo bead](https://redfish.acequia.io/guerin/.agents/cef6a5a8-ccbc-4102-bf59-4c37c82c29f5/about.md), 🔒). The Maria cut has **the most complete story with no narration**; the [Adams narration](https://redfish.acequia.io/guerin/.agents/1f914c9d-1d55-4dcc-9ba5-1d0d921e0502/2026-07-06/artifacts/adams-fire-narration-transcript.md) is the voice template. What makes Maria the better vision piece: it shows the **whole citizen-up loop**, and the narration should make that loop unmistakable.
## Source material | Asset | Where | |---|---| | Maria Fire cut (2:46, silent) | [maria-fire.mp4](https://guerin.acequia.io/videos/maria-fire.mp4) · [youtu.be/aJpgDzFhXng](https://youtu.be/aJpgDzFhXng) · [go.simtable.com/maria-fire](https://go.simtable.com/maria-fire) | | Adams cut (3:30, narrated; style template) | [adams-fire.mp4](https://guerin.acequia.io/realtime-earth/videos/adams-fire.mp4) · [go.simtable.com/adams-fire](https://go.simtable.com/adams-fire) | | Adams narration transcript | [artifacts/adams-fire-narration-transcript.md](https://redfish.acequia.io/guerin/.agents/1f914c9d-1d55-4dcc-9ba5-1d0d921e0502/2026-07-06/artifacts/adams-fire-narration-transcript.md) | | Keyframes (this cut) | [artifacts/keyframes/](https://redfish.acequia.io/guerin/.agents/1f914c9d-1d55-4dcc-9ba5-1d0d921e0502/2026-07-06/artifacts/keyframes/) (maria-NNN.jpg; frame N ≈ (N−1)×4s) | | Landing page | [go.simtable.com/realtime-earth](https://go.simtable.com/realtime-earth) |
## What the video actually shows (timecoded story map) | ~Time | Frames | Beat | |---|---|---| | 0:00–0:08 | 001–002 | **Realtime.Earth Live Incident View**: the Maria Fire spread raster on the basemap (the destination picture, shown first) | | 0:08–0:20 | 003–005 | **Reported Fire Area** ellipse on the map; the AlertWildfire.org camera located relative to it | | 0:20–0:32 | 006–008 | **Calibrated Camera Viewer**; "Camera Capturing Fire Start": the tower camera catches first glow at dusk over the city | | 0:32–0:44 | 009–011 | **Citizen Pilot Footage (Twitter: @acloudaway1)**: a raw aerial photo of the young fire, then placed onto the georectified 3D terrain — **the imagery intel analyst georectifying a raw tweet image, the first image of the fire** | | 0:44–1:00 | 012–015 | Plume visible from city cameras (color + monochrome/IR); "Video with Estimated Perimeter": the first perimeter estimate draped on terrain | | 1:00–1:08 | 016–017 | **LiveMap Alert Posted — the PIO beat**: @VCFD_PIO tweet, "At 6:14PM @VCFD responded to a confirmed brush fire on South Mtn in Santa Paula… Firefighters are working to keep the fire within the following box: S of 126 fwy, N of LA Ave, E of Vineyard, W of Balcom," with an embedded **Realtime.Earth LiveMap** of that box (262 retweets). The alert carries the live URL map | | 1:08–1:24 | 018–021 | **Citizens receive the alert and provide imagery back**: "Citizen Footage" georectified into an **Estimated Fire Perimeter** band on South Mountain; "Citizen Footage (Periscope: @MitchelBrewer)" night panorama | | 1:24–1:32 | 022–023 | **Annotate Calibrated Imagery**: analyst drops annotation points/lines directly on the calibrated night imagery | | 1:32–1:44 | 024–026 | **AVL/GPS augmentation begins**: dozer marker on the 3D terrain; "Compare Data to Perimeter Estimate" (GPS track against the perimeter) | | 1:44–2:04 | 027–031 | Progression playback through the calibrated smoke views; annotations advance; "Continue Playback" | | 2:04–2:20 | 032–036 | **Visualize Perimeters with Calibrated Video**: rainbow isochrones draped on the camera terrain | | 2:24–2:32 | 037–038 | **Rate of Spread Analysis — measuring actual fire behavior**: isochrones plus an inset matching observed rate of spread (acres/minute over time) against **prior fuel raster data** (Fuel Types 1, 2, 5, 8, 9, 93) | | 2:32–2:46 | 039–043 | **View AVL Data on Camera**: Dozer 110 and Dozer 129 (Division Z) and a Type 6 Engine labeled live on the smoke-column camera view; closing vista | Southern California Edison branding appears throughout (partner credit); the closing AVL labels include **crew names and phone numbers** (see edit considerations).
## Why this is the vision piece The cut demonstrates, in one incident, the loop the strategy talks about: a calibrated camera sees first → an analyst georectifies a citizen's raw tweet photo into the county's picture → **the PIO pushes an alert that carries the live map** → citizens who receive it contribute footage back → the analyst annotates and fuses → observed fire behavior is measured and checked against prior fuel data → crews and equipment are tracked in the same picture. Alert out, imagery in, models validated, everyone's view relevant to their seat. The narration's job is to say that loop out loud.
## Instructions for the edit 0. **Brand the narrated cut as alert.live** (Stephen, 2026-07-06). The story this video tells, the VCFD PIO putting out an alert with a living map and citizens answering with imagery, is **alert.live's story**. Realtime.Earth is the foundation technology underneath, less a vertical product; credit it as the platform ("built on Realtime.Earth"), never as the headline. Title card, lower-thirds, and end card carry alert.live; the on-screen 2019 UI watermarks say Realtime.Earth and that is fine as historical footage under an alert.live frame. 1. **Confirm cut points** against the timecode map above (final viewing pass; the keyframes make this fast). 2. **Keep ~2:46; do not pad.** Narration budget ≈ 380–400 words at ~145 wpm. Draft below is ~360. 3. **Language rules** (calpoly-slo guide): community resilience, never climate; no em dashes; no not-this-but-that; concrete Adams-style verbs; one vision sentence at the close. 4. **Crew names/phone numbers in the AVL labels (2:32–2:46):** decide keep vs blur. The footage is already public on YouTube, and the labels demonstrate the capability vividly; still, for a cut we will mail to new partners, consider blurring the phone numbers while keeping unit labels. 5. **Credit line:** keep the Edison partner credit and the ALERTWildfire camera lineage; add citizen contributors' handles as displayed (@acloudaway1, @MitchelBrewer). 6. **End card / CTA:** headline **alert.live**, "built on Realtime.Earth" as the platform credit, destination ⟦Stephen: alert.live landing vs go.simtable.com/realtime-earth vs go.simtable.com/maria-fire — pick the URL the end card names⟧. 7. **Publish path:** narrated master behind [go.simtable.com/maria-fire](https://go.simtable.com/maria-fire); update the learnmore page card; cut a 60–75s short for email/social.
## Draft narration script (fitted to the story map; confirm against final cut points) > **[0:00]** On Halloween evening 2019, the Maria Fire broke out on South Mountain between Santa > Paula and Somis in Ventura County. This is the picture of that fire a community can own: every > frame of it generated directly from cameras. > > **[0:12]** An alertwildfire.org camera sits within view of the reported fire area. In alert.live > that camera is calibrated: every pixel is tied to a latitude and longitude on the ground, so > imagery can be read like a map, and a map can be drawn from imagery. > > **[0:24]** The calibrated camera captures the first glow at dusk. > > **[0:34]** Minutes later, a citizen pilot posts a raw aerial photo to Twitter. An imagery analyst > georectifies that single photo: the first image of the fire becomes the first measurement of the > fire. > > **[0:48]** From city cameras, the plume is visible across the valley, and a first perimeter > estimate is draped on the terrain. > > **[1:00]** The Ventura County Fire PIO posts the alert, and the alert carries a live map. The > containment box the firefighters are working: south of the 126, north of LA Avenue, east of > Vineyard, west of Balcom. Hundreds of residents share it within the hour. > > **[1:10]** Citizens who receive the alert send imagery back. Their footage is georectified into > the estimated fire perimeter, and the picture sharpens with every contribution. > > **[1:24]** Analysts annotate the calibrated imagery directly: what is burning, where the line is. > > **[1:34]** Vehicle locations join the same picture. A dozer's GPS track is compared against the > perimeter estimate as it works. > > **[1:46]** Through the night, playback of the calibrated cameras records the fire's progression. > > **[2:06]** Perimeter isochrones are drawn from the imagery, hour by hour. > > **[2:24]** And the fire's actual behavior is measured: observed rate of spread, matched against > the fuel data on the ground before the fire. Real observation, checking real models. > > **[2:34]** Dozers and engines are labeled live in the camera view, so the people managing the > incident see resources and fire in one picture. > > **[2:42]** A community is blind in the first hours of an emergency, even while its own cameras > and phones are watching. alert.live ends that blindness: alerts go out with a living map, imagery > comes back from the people, and the community sees itself and acts together. alert.live, built on > Realtime.Earth. ⟦end-card URL per instruction 6⟧
## The vanguard extension (for the calpoly-slo engagement) Stephen (2026-07-06): **alert.live is an onboarding**, and SLO can go further: **train its citizens to be the vanguard, and a trained citizen can start an incident.** The Maria loop began with a tower camera and a lucky pilot's tweet; a trained vanguard makes that first-report loop deliberate. Residents who register for alerts, know their zone (FireSafeSLO's Know Your Zone channel), and have practiced contributing calibrated imagery become the community's first sensors: the first report, the first photo, the first georectified measurement can all come from a citizen. For the narrated cut this stays as the close's implication; for the SLO partnership it is a program to propose (vanguard training through the Firesafe Council's education channel).
## Done means Narrated master + 60–75s short published behind [go.simtable.com/maria-fire](https://go.simtable.com/maria-fire), linked from the Frievalt reply's "newer to our work" block, with captions, credits, and the AVL-label privacy decision made.
## References (bead cross-links) - Bead cef6a5a8 · [canonical](https://redfish.acequia.io/guerin/.agents/cef6a5a8-ccbc-4102-bf59-4c37c82c29f5/) (no page yet)