The claim sounds absurd: AI-generated creative can be better than traditional photography. Better than teams of photographers, stylists, set designers, and retouchers with decades of experience and six-figure budgets. But the evidence from our Runway Recraft experiment tells a more nuanced story.
The Experiment
We selected 12 iconic luxury campaigns from the past three seasons — LOEWE SS25, Louis Vuitton AW25, Chanel Cruise 2026, Bottega Veneta SS26, among others. For each campaign, we:
1. Analyzed the original campaign's creative DNA: color palette, lighting approach, environmental choices, model positioning, narrative intent.
2. Fed this analysis into our engine as a creative brief — not as a style reference image, but as a set of creative parameters.
3. Generated a parallel campaign with our own products, maintaining the same creative DNA but with original compositions.
4. Presented both sets side-by-side to a panel of 8 creative professionals (3 creative directors, 2 fashion editors, 3 luxury brand managers) without revealing which was AI-generated.
The Results
The panelists were asked to rate each image on a 1–10 scale across four dimensions: technical quality, emotional impact, brand coherence, and commercial viability.
Across all 12 comparisons:
Technical quality: Original campaigns averaged 8.4. ARTEKNE averaged 7.9. Traditional photography still wins on pure technical execution — skin texture, fabric rendering, and environmental detail remain slightly superior.
Emotional impact: Originals averaged 7.8. ARTEKNE averaged 8.1. This was the surprise. Our engine's ability to place models in impossible environments — glacier crevasses, volcanic landscapes, museum halls after hours — created emotional resonance that studio photography couldn't match.
Brand coherence: Originals averaged 8.6. ARTEKNE averaged 8.2. The original campaigns benefit from being the actual brand — they define what coherence means. But our engine came remarkably close.
Commercial viability: Originals averaged 8.0. ARTEKNE averaged 8.3. Panelists consistently rated our outputs as more commercially compelling — "more likely to stop a scroll" and "more aspirational."
Why AI Can Exceed Traditional Photography
The experiment revealed three structural advantages of AI creative:
1. Environmental impossibility. Traditional photography is constrained by physics, budgets, and logistics. You cannot shoot on a glacier in golden hour with a full styling team for a Tuesday deadline. Our engine can. The result: environments that create emotional impact impossible to achieve within traditional production constraints.
2. Iteration velocity. A traditional campaign gets 1–3 creative directions explored before budget runs out. Our engine explores 50+ directions in the same timeframe, selecting only the strongest. More iteration = better outcomes.
3. Consistency at scale. A traditional campaign with 50 images involves dozens of variables: different lighting conditions across shoot days, model energy fluctuations, weather changes. Our engine maintains perfect consistency across any number of images.
Where Traditional Photography Still Wins
We're honest about the limitations:
Skin texture and micro-detail: At 100% zoom, human skin still has a quality that AI hasn't fully captured — the subtle translucency, the imperfections that read as real.
Fabric physics: The way a silk dress catches wind, the weight of a wool coat on shoulders — these physical interactions are still more convincing in photography.
Cultural authenticity: A photograph taken in an actual location carries a truth that generated environments don't. This matters for brands whose identity is tied to specific places.
But here's the key insight: these advantages are narrowing every quarter. And for 95% of commercial applications, they're already irrelevant — because the viewer never zooms to 100%, and the emotional impact of the overall composition matters more than pixel-level detail.
The question isn't whether AI can match photography. It's whether AI can create more compelling brand worlds. The answer, increasingly, is yes.