AI & Visual
Virtual Try-On in Fashion
Seeing a product on yourself before buying is a compelling promise, held back by the stubborn complexity of bodies, fit, realism and privacy.
Updated 2026-07-01 · 7 min read
Returns are the tax that online fashion pays for uncertainty, and much of that uncertainty is about how something will actually look and fit. Virtual try-on aims straight at it: let shoppers see a product on themselves, or a close proxy, before they commit. The pitch is obvious; the execution is not.
Try-on ranges from simple overlays to full body-aware simulation, and the harder the fashion problem, the more the technology strains. This page is deliberately even-handed about a much-hyped area, weighing genuine use cases against real limits in body variety, sizing, realism and privacy. It ties into size and fit guidance and the wider arc of the future of fashion search.
The use cases that work today
Try-on is most convincing where the problem is constrained. Eyewear, watches and some footwear sit in fixed positions relative to the body, so augmented-reality overlays can look plausible and genuinely help a shopper judge scale and style. For accessories and cosmetics, adoption is well established and the value is clear.
Even simple try-on can reduce hesitation and, in principle, returns, by replacing pure imagination with a concrete preview. Where it works, it works because the underlying geometry is simple and forgiving, not because the fashion problem has been solved.
Why bodies make it hard
Garments are where try-on gets difficult, and the reason is human variety. People differ across height, weight, proportion, posture and shape in ways no single avatar captures, so a garment rendered on a standard model tells a specific shopper very little about their own experience of it. The more the body diverges from the template, the less the preview means.
This is also a fairness issue. Systems trained and demonstrated mostly on a narrow range of body types serve those bodies well and everyone else poorly, quietly signalling who the experience is for. Honest try-on has to represent the full range of real bodies, which is far harder than rendering an idealised one.
Sizing complexity compounds the problem
Even a perfect visual does not answer the size question. A garment that looks right draped on an avatar may still fit badly because fit depends on measurements, cut and fabric behaviour that appearance alone does not convey. Visual try-on and sizing are related but distinct, and conflating them sets shoppers up for disappointment.
Fabric adds another layer: the same size behaves differently in rigid denim, stretch jersey or bias-cut silk, and simulating drape convincingly is an unsolved problem at scale. This is why we treat sizing as its own subject rather than a footnote to try-on. Looking right and fitting right are not the same claim.
The realism ceiling
Realism cuts both ways. A preview that is too crude is unpersuasive and ignored; one that is too polished can flatter, showing a garment behaving better than it will in reality and setting expectations the product cannot meet — which risks more returns, not fewer. The credible target is accuracy, not beauty.
Rendering fabric physics, lighting and fit convincingly in real time remains computationally hard. Many deployments settle for a stylised approximation, which is honest about its limits but less compelling than the marketing implies. The gap between demo and daily use is still wide.
Privacy and consent
Try-on frequently needs the shopper's body: a photo, a scan, or measurements. That is among the most sensitive data a retailer can hold, and it raises pressing questions — where images are processed, whether they are stored, how long they are kept and whether they train future models.
On-device processing that never uploads a body image is far more privacy-preserving than server-side capture, and shoppers are increasingly discerning about the difference. Any try-on programme has to make consent and data handling explicit; treating body data casually is both an ethical failure and a business risk, a theme that connects to responsible data use across the future of fashion search.
Where the potential is real
The honest assessment is that virtual try-on is promising and immature. It clearly helps for constrained categories and struggles with the full complexity of garment fit across diverse bodies. Progress will come from better body representation, more accurate fabric simulation and privacy-first architectures — not from prettier renders.
Business of Fashion and McKinsey have both flagged try-on as a watched area with adoption still uneven across categories. Track it as an evolving capability, not a finished one, and see how it fits the wider field in our fashion-tech trends overview and the discovery hub.
Sources & further reading
- Business of Fashion, “The State of Fashion” (2024)
- McKinsey, “The State of Fashion” (2024)
Sources are attributed to their publishers and link to each publisher's own site. Figures reflect general market direction rather than point-in-time precision; consult the linked publishers for their current data.