Real-Time Virtual Try-On: What It Is and Whether Your Store Needs It Yet

Searches for virtual try-on increasingly turn up demos where the garment moves with a shopper's own camera feed rather than sitting still in a generated photo. This is a new capability, not a rebrand of AR filters or a faster version of image generation. Here is what it actually is, who is building it, and what it means for a store deciding what to add to a product page today.

What is real-time virtual try-on?

Real-time virtual try-on streams a shopper's own live camera feed back to them with a garment rendered onto their moving body, continuously, instead of producing one still image. Turn to the side and the jacket turns with you. Walk and a dress moves the way it will on the street. The clothing updates at video frame rates instead of being calculated once and shown as a finished photo.

That sets it apart from both categories covered in AI virtual try-on vs AR filters. AR filters track a face or wrist and pin a simple object onto it live, but were never built to simulate how fabric drapes across a moving body. Standard AI virtual try-on, the kind Corlen and most competitors run, generates a new photo from a single upload and takes several seconds to render fabric and fit properly. Real-time try-on aims for the fidelity of the second category at the speed of the first, and doing that for a full outfit is harder than either one alone.

Who is actually building real-time virtual try-on?

Two examples shipped within days of each other in August 2026, aimed at different parts of the market.

• Remark: launched a real-time virtual try-on experience on August 20, 2026, built for luxury retail. It runs directly from a brand's product page with no app or download, rendering the garment on a shopper's moving body in full HD at real-time frame rates, and lets a shopper swap sizes, colors, and full looks mid-session. More than 70 brands were on it at launch, including Ariat, J.McLaughlin, and Veronica Beard, rolling out through fall 2026.
• Decart: ships Lucy 2, a developer-facing model that takes a live camera feed plus a garment reference image and streams back video of the person wearing it, with near-zero latency and no per-frame server round trip. It is available through Decart's own API and through fal.ai, the same infrastructure layer Corlen's own try-on engine ran on in an earlier version, before the engine moved to Google's Gemini image model directly.

Both are aimed at different buyers. Remark sells a finished, brand-facing experience to luxury retailers. Decart sells the underlying model to developers building their own experience. Neither is a general-purpose tool a merchant installs the way a Shopify app installs in a few clicks today.

How is this different from generating a try-on image?

The input is the same basic idea in both cases: a shopper's likeness and a garment. What changes is what happens after that.

What are the tradeoffs right now?

Neither approach is strictly better than the other today. Each comes with its own limits.

• A live camera has to stay open: real-time try-on needs continuous camera access, which rules out any flow where a shopper previews a look without a device pointed at them in the moment, including email or WhatsApp.
• Lighting and camera quality matter more: a generated image can come from one decent photo taken once, under whatever light was available at the time. A live feed has to hold up for the whole session, and a dim room or a shaky hold degrades the result in a way a single well-lit photo does not.
• Reach is narrower today: Remark's rollout covers more than 70 luxury brands, not an open install any store can add. Decart's model is a developer building block, so using it means building the product-page experience from scratch rather than installing an app.
• Nothing is left to share afterward: a generated try-on image saves as a file a shopper can send to a friend for a second opinion. A live camera session ends the moment the tab closes, with nothing left to forward.

Should a fashion store wait for real-time try-on?

For nearly every store outside Remark's luxury brand list, no. Image-based try-on is the only version of this a merchant can add to a product page today without building custom camera infrastructure or waiting on a partnership.

The two categories are not competing for the same purchase decision yet. A shopper deciding whether a jacket fits, in a size they picked, from a photo they already have, is well served by a generated image in 10 to 15 seconds. Real-time try-on solves a different problem: an in-the-moment, camera-open experience closer to a mirror than a product photo, a stronger fit for a luxury flagship page than a store's regular product grid, at least at this stage of the technology.

Corlen's own approach stays image-based: a shopper uploads or captures one photo, and the same engine behind the kiosk and the developer API generates a result in seconds, with no live camera required and no dedicated hardware. Whether that changes depends on how soon real-time rendering gets fast enough, cheap enough, and easy enough to install the way an app installs today, not on whether the underlying idea is sound.

See Corlen's current approach on your own photo: try it for free, or read the developer docs to see the response format behind it.