Decart AI's Lucy 2.0: The Real-Time Video Revolution

Decart AI's Lucy 2.0 made waves by demonstrating real-time video generation fast enough to change a person's outfit live, frame by frame. It's a genuine research milestone, and a useful lens on where virtual try-on is heading and what actually matters for merchants today.

What Lucy 2.0 Actually Showed

Lucy-class models generate video continuously at interactive framerates, meaning a camera feed can be restyled with different clothes and different environments with no per-frame wait. The demos are striking: outfits that move with the person in real time rather than being stitched onto a still.

The Gap Between Demo and Checkout

Real-time video try-on still fights three battles before it's a commerce tool: garment fidelity (a merchant needs their exact SKU rendered faithfully, not a plausible jacket), consistency (flicker destroys buying confidence), and cost per session at retail traffic volumes. Photorealistic still-image try-on, where physics simulation preserves the actual product's drape and print, is what converts shoppers today.

What Merchants Should Take From It

The direction is confirmed: seeing products on yourself is becoming the default expectation, and the technology curve is steep. The practical move is to adopt the version that works now (about 9 seconds per still-image render with DrapX) so your store and data are ready as video try-on matures.