I built an AI video generator for a home builder client over the weekend. The following social-ad ready video was generated using a readily available image that the client had in their library.

Here’s the problem. As a Digital Media Lead for one of Calgary’s biggest home builders, one of the biggest bottlenecks in our campaign workflow is rotating video assets. Especially on paid social.

So I used Claude to build a workflow that takes readily available listing photos and turns them into ad-ready video assets at multiple resolutions. At scale.
→ No production crew needed
→ No weeks-long creative turnaround
→ Multiple aspect ratios for every paid social placement
→ Built from photos the client already has on hand

The output isn’t cinematic. But for paid social where you need fresh creative rotating constantly to fight ad fatigue? It gets the job done.

Below is the process I followed.

The tools: Node.js (free runtime), Runway Gen-4 Turbo API (AI video generation at $0.50 per clip), and FFmpeg (free open-source video processing).

The process:

I got Claude to write two Node.js scripts. The first takes a listing photo and sends it to Runway’s Gen-4 Turbo API along with a carefully engineered prompt, something like “slow cinematic camera pan across this room, warm natural light, professional real estate videography style.” Runway’s AI analyzes the actual image and generates a 10-second video with realistic camera motion. The key here is that the video preserves the real room. It’s not generating a fake kitchen; it’s animating the actual listing photo with cinematic movement.

The second script takes that raw AI video and runs it through FFmpeg to add branded text overlays (property price, bed/bath count, call-to-action) and exports it in two formats: 9:16 for Reels and Stories, and 1:1 for Feed ads. The text sits in semi-transparent dark bars at the top and bottom so it’s readable without obscuring the property.

Everything runs from PowerShell on my Windows laptop. I drop listing photos into a folder, run two commands, and get back finished ad creatives. No editing software, no design skills, no waiting on freelancers.

What it costs: $0.50 per video in API credits. The branding step is free since FFmpeg runs locally. For a typical builder community launch with 10 properties and 2 videos each, that’s $10 total compared to $2,000+ for a videographer.

The four video styles I built prompts for: interior pan (smooth horizontal movement across a room), exterior dolly (gentle push-in toward the home’s front entrance), walkthrough (forward-moving camera as if walking into the space), and lifestyle (slow cinematic drift with shallow depth of field for aspirational hero shots).

The workflow supports batch processing. Point it at a folder of images and it generates videos for all of them sequentially. This is what makes it practical at scale. A community launch with 50 lots and 3 photos each produces 150 ad-ready videos for $75.

In 2026, creative bottlenecks shouldn’t be the reason your campaigns underperform. The tools exist. The barrier is knowing how to connect them.