Generative video try-on: motion and drape
The logical extension of static photo try-on is a short video clip showing the shopper wearing the garment in motion — walking, turning, or gesturing. Video allows shoppers to evaluate how fabric moves, how a hem falls when walking, and how structured garments maintain their shape under dynamic conditions. Research groups at several AI labs demonstrated early versions of garment-on-video transfer in 2024 and 2025, with quality improving rapidly.
The commercial threshold for video try-on requires temporal consistency — the garment must stay correctly rendered across every frame without flickering or warping artifacts — which is a significantly harder problem than single-frame rendering. Generating a 3-second clip at acceptable quality currently takes minutes on high-end hardware, versus 8–15 seconds for a single image. Plausible estimate for commercial-grade apparel video try-on at acceptable latency: 2028.
Live AR meets generative AI
Current AR try-on (real-time camera overlay) and current AI try-on (render from a static photo) are separate technology stacks. The next synthesis is a live camera feed processed by a generative model in near-real-time — eliminating the 'point your camera and see a stiff 3D overlay' limitation of AR while preserving the immediacy of a live experience. Early demonstrations exist as research prototypes, typically running at 2–5 frames per second on mobile hardware as of 2025.
Achieving the 30+ fps required for a natural live try-on experience requires either specialized inference hardware (unlikely to be standard in consumer devices before 2028) or aggressive model compression research. This is a plausible medium-term development but should not be presented as imminent. The near-term value for merchants remains in static photo-based rendering, which already delivers the conversion outcomes that matter.
Biometric-free fit prediction
One of the persistent gaps in virtual try-on is that it can show how a garment looks but not how it fits — whether it will be too tight at the waist, too long in the sleeve, or too short in the torso for a specific body. Fit prediction requires body measurements, which current systems obtain either through user self-reporting (inaccurate) or through 3D body scanning (unavailable to most online shoppers).
Research on inferring body measurements from a single 2D photo — using silhouette analysis and pose estimation — has made meaningful progress. Systems that can estimate a shopper's approximate measurements from a selfie with 2–3 cm accuracy across key dimensions are commercially realistic in the 2027–2029 window. When combined with structured garment measurement data from brands, this would enable genuine fit prediction without requiring a tape measure or specialized hardware.
Multi-garment outfit composition
Current photo-based try-on handles one garment at a time. A shopper can see themselves in a specific dress or a specific jacket, but not both together with an accessory. Full outfit composition — simultaneously rendering a top, bottom, layer, and accessory on the same photo — requires solving garment-garment occlusion and interaction, which is substantially more complex than single-garment rendering.
Early commercial implementations of multi-garment composition appeared in 2025–2026 for simpler combinations (top plus bottom, dress plus accessory). Full stack outfit rendering at photorealistic quality is a 2027–2028 development. For fashion merchants, this feature is most valuable for stores that sell coordinated sets or have a strong 'shop the look' purchase pattern, where seeing a complete outfit can raise AOV by 30–50% compared to single-item purchases.
Persistent consumer body models: the medium-term platform shift
The most commercially significant medium-term development is the persistent body model: a digital representation of a shopper's body that they build once and reuse across multiple shopping sessions and multiple retailers. Instead of uploading a new photo every time, the shopper's body model is stored (with their consent) and serves as the base for every try-on. This dramatically lowers the friction of the try-on experience and enables cross-retailer fit consistency.
The business model implications are significant. The entity that holds a consumer's persistent body model has a distribution advantage across every retailer that integrates with the platform. This is a winner-take-most dynamic, and it is not yet clear which player will occupy that position — the device manufacturer, the operating system, a dedicated fashion platform, or one of the major ecommerce platforms. For now, this is a strategic horizon item rather than an operational one. What merchants should act on today is deploying the photo-based try-on that already delivers proven ROI.