Leaf – 3D Realistic Cup & Drinkware

Leaf Drinkware had a timing problem disguised as a creative problem. A beautifully engineered cup, a confirmed Amazon launch slot, and an influencer seeding plan were all pointed at a single launch date that was four weeks out. The traditional path — a professional product photographer, retouching, and a 3D agency for lifestyle renders — came back at a fifteen-thousand-dollar quote and a six-week production window. That was not a negotiation; it was a missed launch. Worse, their samples were still finalizing in colorway and handle geometry, meaning any conventional shoot would lock them into expensive reshoots every time engineering made a late change. We rebuilt the entire product-visualization stack as an AI-first pipeline, with the goal of letting the engineering team keep iterating right up to the launch date without collapsing the marketing timeline. The work started with CAD intake. Geometry was converted into image-to-image references using Flux and Midjourney, anchored to studio lighting prompts we built and tested specifically for physically-based rendering accuracy on coated stainless steel and frosted glass — the two finishes on the product line. From that foundation we generated hero packshots in eight angles across three colorways, running every image through an 8K upscale and a Photoshop composite pass to add the details AI does not reliably produce: condensation droplets, steam curl, subtle logo debossing, micro-scratches that communicate real use. Lifestyle scenes were generated in parallel using real-world context prompts — a morning kitchen, a hiking trail at blue hour, a Tokyo café — and staged to match the brand without a single location scout. The packaging dieline was dressed in final imagery, the Amazon PDP stack was delivered at spec for every slot, and a looping hero video for the landing page was rendered in Runway from one of the lifestyle scenes. When engineering revised the handle radius mid-project, we regenerated the full hero set overnight using the saved prompts, which would have been a six-figure problem on a traditional shoot. The complete visual library — forty-five-plus assets — shipped in two weeks at roughly ten percent of the traditional cost. Launch week pre-orders came in at $180k against a $95k plan, and the team now has a prompt library they can re-run every time they add a colorway without commissioning another shoot.

More from this series lives in our AI creatives gallery, alongside campaign reels and brand experiments.