How we delivered a full wet food campaign from two existing dry food commercials, using AI-powered production.
Affinity needed new TV spots for their Natural Trainer wet cat food line. They already had two professionally produced commercials for their dry food range and wanted to extend the campaign to a different product category while maintaining full brand consistency. The goal was clear: deliver broadcast-quality spots for wet food without a single additional day of filming.
This was not a simple edit or recut. The brief required fundamentally different creative elements across every dimension of the spots:
New people. The actors from the original spots needed to be replaced entirely with new, AI-generated talent.
New product. Every pack shot and product interaction had to shift from dry food (large bags, kibble) to wet food (smaller packs, poured servings). The product had to look physically accurate.
New environments. The spots needed additional rooms and angles within the same house setting that were never filmed during the original production.
New animal behaviour. Both adult cats and kittens were required. Filming kittens is notoriously difficult in traditional production: they don’t repeat takes, they react unpredictably to lights and equipment, and a full day on set might yield only a few usable seconds.
All of this had to result in finished, broadcast-quality commercials ready for public release. Not a proof of concept. Not an internal test.
Starting from real footage. We used the two existing dry food spots as our creative and visual foundation. The lighting, framing, colour grading, and overall production language served as the reference for everything we generated.
Expanding the environments. Using the original filmed locations as a base, we generated additional rooms and perspectives within the same house. The AI extended the physical space while maintaining full visual coherence with the real footage.
Generating new talent. We replaced the original actors with entirely new, AI-generated people. These had to match the tone, styling, and demographic profile that the brand required for the wet food campaign.
Training the AI on the real product. This was critical. Affinity provided us with physical product samples, and we shot reference photos and videos of the actual wet food packs and the product being served onto plates. We tested angles, textures, and pouring behaviour with real footage, then used that reference material to train the AI so that every generated product shot was faithful to the real thing. We did not invent the product visuals. We derived them from reality.
Creating animal motion. We generated realistic cat movements for both adults and kittens. This required extensive iteration, as the AI’s initial interpretation of a kitten’s movement was closer to that of a small wild feline than a domestic cat (more on this in Learnings).
Assembly and delivery. The final spots were assembled, graded, and delivered in two durations each, resulting in four finished versions ready for broadcast and digital distribution.
4 finished spots delivered (2 spots, each in 2 durations), fully produced using AI from existing footage.
1 month of production time, compared to an estimated 2 months for a traditional reshoot including pre-production and post.
Less than half the cost of what a conventional production would have required.
Published and live on YouTube. These are real, public-facing campaign assets for a major petcare brand.
This project proves that hybrid and fully AI-powered production can deliver broadcast-standard advertising from existing material. One shoot becomes the foundation for multiple campaigns across different products, markets, and creative directions.
AI interprets animals differently than we expect. When generating kitten motion, the AI consistently produced movements closer to a small wild cat (a lion cub, a jungle feline) than a domestic kitten. This taught us that guiding AI animal behaviour requires very specific reference and iteration. The gap between “small cat” and “kitten” is something the models don’t yet distinguish intuitively.
Real reference material is non-negotiable for product accuracy. Generating product shots without physical reference produces plausible but inaccurate results. The step of shooting real product footage and using it as training material made the difference between “looks like cat food” and “looks exactly like Natural Trainer wet food.”
Client courage matters. Affinity committed to a fully AI-generated campaign for a major product line. That decision required trust in the process and in the production partner. Not every client is ready for this step yet, but the results demonstrate that the quality and reliability are there.
One shoot, unlimited potential. This case validates what we believe is the future of advertising production: a single, well-executed shoot becomes the seed for an entire system of campaign variants. Different products, different markets, different targets, different creative executions. All without returning to set.
Make it real.
