Generative AI

Oh no, AI.

AI is changing the way creative work gets made, and we don't shy away from that. We embrace it. But we also believe the difference isn't the software, it's the creative direction, taste, and technical experience behind it.

Our background in design, CG, VFX, animation, and post allows us to use AI with intention, combining new tools with proven workflows to create work that's faster, more flexible, and more art directed.

Client
Warm & Fuzzy
Services
  1. AI & Emerging Tech
  2. Visual Effects
  3. Editorial & Finishing
Year
2026

Explore

13 Assets

Restoring the past

Hershey’s wanted to bring back a 15-year-old KitKat Halloween commercial with updated packaging. The only source was a low-resolution 360p MP4. We used upscaling and decompression models tailored to each shot, then moved into traditional tracking and compositing to replace the packs. An old spot, ready for another Halloween.

Enhancing CG renders

Water droplets, frost, and subtle surface details can take a long time to build in CG. Our hybrid approach keeps the camera, animation, and lighting in a traditional pipeline, then uses AI for a final detail pass. The La Colombe comparison shows how condensation changes the finish while preserving the underlying composition.

  1. La Colombe vanilla latte can before and after AI adds condensation.

Hybrid VFX workflows

For Dunkin’, we used AI to generate backgrounds and help with tracking and keying, including tricky foliage edges. Before processing Rachel Dratch’s shot, we obscured her face, then composited the original performance back into the result. We also developed a tool to restore missing frames and match the source timing.

In the car-interior shot, AI helped us relight the scene and change design details. We then handled the background replacement with traditional compositing. Each part of the shot went through the workflow that suited it.

Creative R&D

Prompt-based exploration lets us move from a loose reference to a specific image quickly. We use it to prototype ideas, explore visual directions, and help clients picture a concept. These experiments range from product worlds and cinematic environments to portraits, motion tests, and a rider flying the Warm & Fuzzy flag.

  1. Eight concept images exploring riders, racing, flowers, portraits, and a cat’s eye.
  2. Monster Energy concept triptych with black boots, a can on concrete, and a woman holding a can.
  3. Rear view of a car with exposed bodywork on a wet mountain road.
    Finished silver sports car on a wet mountain road.

Custom tools

AI has also accelerated how we build tools for our own pipeline. Our custom Premiere plugin automates much of the work of updating graded footage and VFX versions across multiple edits. Less time replacing shots by hand, more time working on the edit.

  1. Node-based image workflow connecting product references and generated concepts.

Dunkin Pumpkin Spice