A magic-like wardrobe change powered by generative AI image synthesis. The solution fuses the 2D product photo from the EC site with the user's own image. No 3D garment data needed, which means the entire existing inventory a brand already owns becomes try-on-ready content starting today.
まるで魔法のように服を着せ替える、AI画像生成型バーチャル試着。 最新の生成AI（Generative AI）技術を活用し、ECサイト等にある「商品の2D写真」と「ユーザーの画像」を合成するソリューションです。 服の3Dデータを一から作る必要がないため、ブランドが抱える膨大な既存在庫を、 今日からすぐに「試着可能なコンテンツ」へと変えることができます。
To avoid the pattern distortion and quality instability typical of on-the-fly generative AI, we adopt a two-stage pipeline — pre-training separated from instant display (patent pending). Distinctive patterns and fine details come through stably, at high fidelity.
Even as advanced AI, we've driven compute cost down below existing solutions. Large EC sites with heavy concurrent load and low-end smartphones both run smoothly.
利用に必要な条件 — 驚くほどシンプル
No special 3D pattern data needed. Any existing EC-site or catalogue 2D image — product-only or on a model — will work.
An avatar image from the VRC platform, or a front-facing photo taken by the user.
While browsing an EC site — or by scanning the QR code on an in-store tag — the shopper selects the garment they want to try.
The user image and the selected product's 2D image are sent to VRC's lightweight cloud AI engine.
VRC's proprietary algorithm composites the image — the user wearing the garment, beautifully dressed, appears on screen.
Without waiting, the shopper swaps colours and tries other items with a tap — iterating freely, then pressing purchase with confidence.
This is pixel-level image synthesis. Fabric stretch and pressure distribution under the arm (Pressure Map) — physical simulations of that kind are out of scope.
Results depend heavily on the angle of the provided product image — the full 3D freedom of 3D VTO (rotating to see the back of the garment, for instance) is not available.