Gift Yourself · Nissan Holiday Retail 2026 · Case study
A frame-accurate, drag-to-compare tool built to evaluate winter visual treatments for Gift Yourself, a 2026 Nissan Rogue holiday spot. Each test holds two versions of the same clip — the untreated plate and a winterized pass — in a single interactive frame. Drag the slider, or use the arrow keys, and the split moves in real time, always showing the exact same frame on both sides, because under the hood there's only one video actually playing.
Gift Yourself hinges on snow — the garage wakes under it, the Rogue pulls away through it, the house sits wrapped in it. But the footage available before a shoot date was locked wasn't shot in those conditions: supplied reference plates from an urban street, a residential neighborhood and a rural stretch of road, captured in whatever season and light they happened to be captured in.
Before committing to a shooting plan or a VFX budget, the team needed an honest answer to one question — does a winter treatment actually hold up, shot for shot, across different environments? Rather than judging that from written notes or a handful of isolated before/after stills, this tool let the director, the agency and the post team scrub through full motion side by side and decide together.
It was built twice: once in pre-production against the three supplied plates, and again — in a much rougher pass — directly on footage from the actual shoot, as a fast gut-check on how the real material would respond.
Each comparison is secretly a single video file. The plate and the treated version are rendered side by side into one frame, so there's only one video ever decoding, and the two halves are mathematically guaranteed to be the same frame, every frame — no timecode matching, no drift, no sync logic to maintain.
A canvas sits on top of that hidden video and redraws two crops of whatever frame is currently playing — the left crop from the plate half, the right crop from the treated half — joined wherever the slider sits. Dragging just moves that split point; the redraw happens on every decoded frame via requestVideoFrameCallback, so the comparison always tracks what's actually playing, not a static image.
Tests 01–03 are rough visualizations built from supplied footage, from pre-production. Tests 04–06 are a very preliminary pass on footage from the actual shoot — just a quick look at how it reads, not a finished treatment. Drag any frame to compare — left is the plate, right is the treatment.