STEVE TAPIA
AI PORTFOLIO
MY BLOG
Cackler Lane: A One-Minute Flythrough Built From Listing Photos
I work from home in Clearwater. Cackler Lane was not down the street. It was a manufactured-home listing in California that had been sitting a year, and the dealer sent me the same thing every agent already has: plain photos.
The finished flythrough is 69 seconds. No crew. No site visit. The companion video is the honest version of the same job: original listing still in the corner, AI treatment playing next to it, same house the whole time.
That split screen is the brief. The stills are gray and flat. The after has warm light in the windows, plants, a car in the carport, and the house looking like someone lives there. The flythrough itself is not a drone reel of a mansion. It’s a lived-in minute — a couple at breakfast, music, the home and the park amenities treated like a place instead of a vacant unit.
What actually took the time was not typing a prompt.
It was deciding what “lived-in” meant for a house that had not moved in twelve months. Which rooms to open. Where a couple sitting down to breakfast does more work than another empty kitchen. How far to push sunset and staging before it stops being the listing. What to do with the park, not just the unit. Then cutting it so an agent can post it without a four-minute tour nobody finishes.
I didn’t invent a pipeline for this. The stack on my AIFD profile is the one I ran: structure the job first (script + Milanote), generate in Kling 3.0 and NanoBanana Pro, score in Suno, finish picture and sound in DaVinci Resolve. A few days, not a production schedule.
They posted it. In the first 24 hours it got the highest view count that listing had ever seen. That’s their number, not a trophy. The useful fact is the input: stills. If you have photos, you have a starting point.
A one-minute flythrough will not sell a bad house. It will stop a decent house from looking empty.
The finished flythrough is 69 seconds. No crew. No site visit. The companion video is the honest version of the same job: original listing still in the corner, AI treatment playing next to it, same house the whole time.
That split screen is the brief. The stills are gray and flat. The after has warm light in the windows, plants, a car in the carport, and the house looking like someone lives there. The flythrough itself is not a drone reel of a mansion. It’s a lived-in minute — a couple at breakfast, music, the home and the park amenities treated like a place instead of a vacant unit.
What actually took the time was not typing a prompt.
It was deciding what “lived-in” meant for a house that had not moved in twelve months. Which rooms to open. Where a couple sitting down to breakfast does more work than another empty kitchen. How far to push sunset and staging before it stops being the listing. What to do with the park, not just the unit. Then cutting it so an agent can post it without a four-minute tour nobody finishes.
I didn’t invent a pipeline for this. The stack on my AIFD profile is the one I ran: structure the job first (script + Milanote), generate in Kling 3.0 and NanoBanana Pro, score in Suno, finish picture and sound in DaVinci Resolve. A few days, not a production schedule.
They posted it. In the first 24 hours it got the highest view count that listing had ever seen. That’s their number, not a trophy. The useful fact is the input: stills. If you have photos, you have a starting point.
A one-minute flythrough will not sell a bad house. It will stop a decent house from looking empty.