AI Video Ad Campaign for Beginnings Recovery OKC

Beginnings Recovery OKC needed Meta video ads to drive admissions calls. Addiction treatment sits under Meta Special Ad Category. Demographic targeting is restricted. The creative does all the work.

Generic recovery ads get scrolled past. This audience has seen stock footage of sad people becoming happy people. They have heard the inspirational music. They scroll.

The brief was simple. Build something real enough to stop the scroll and move someone to pick up the phone.

Two films. Two emotional entry points.

I led with story. Research consistently shows emotional resonance drives action on Meta far more effectively than information. So I earned the call to action by making the viewer feel something first.

A semi-realistic 3D animated short following Caleb, a man in his early 30s, from 2am at rock bottom to walking through the doors of Beginnings Recovery at sunrise. No voiceover or any dialogue.

The color arc moves from cold desaturated blues in the first 20 seconds to full warm golden light by the end. The viewer's nervous system follows that arc. By the time the phone number appears they have already felt what choosing help feels like.

Animation creates just enough emotional distance for someone in denial to recognize themselves without the defensiveness that live action triggers. They watch Caleb. They feel Caleb. They do not feel confronted.

A photorealistic short told by an 11-year-old girl about her mother's addiction and recovery. She speaks directly to camera at the end.

This concept reaches two audiences at once. The family member searching for help on behalf of someone they love. And the person struggling, approached from the one angle they cannot defend against. Their child's perspective.

The hook lands in under 3 seconds. A stat card follows immediately so the right audience knows what they are watching and stays.

Built entirely with AI tools. No camera. No crew. No studio.

Nanobana 2 handled all image generation across both concepts. Nanobana 2 was used for animated character generation, storyboarding, and environment references throughout Concept 1 and handled photorealistic character and scene generation for Concept 2.

Runway and Flora handled all video generation and motion work. Seedance 2.0 inside flora was the primary animation engine across both concepts, using Multi-reference and Keyframe generation modes depending on the complexity of each shot. Runway handled video refinement and motion control where precision was needed.

ElevenLabs v3 generated the voiceover for Concept 2 with precise emotional audio tags written specifically for a natural 11-year-old delivery. Suno composed the original score for both films. Everything was assembled, color graded, and finished in CapCut.

Character consistency across 22 scenes in Concept 2 was achieved by generating empty environment references first, then placing consistent characters into locked rooms. Same methodology used in high-end AI narrative production workflows in 2026.

6 total video assets across two concepts. 16:9 and 9:16 versions of each. Full length and 60 second cuts. All Facebook policy compliant. Special Ad Category ready.

Production timelines that used to take weeks and cost tens of thousands were compressed into days. The creative quality did not suffer. It improved. Every frame, every cut, every lighting choice was made with intention rather than budget constraint.

This is what AI video production looks like when it is treated as filmmaking rather than content generation.