Google is extending its AI powered virtual try-on feature with personalised avatars.
Google has introduced a new AI powered virtual try-on (VTO) feature within its AI Mode shopping experience. For now limited to participants in its experimental Search Labs initiative, the tool enables users to upload a full body photograph in order to visualize its apparel listings on their own likeness.
While Google’s generative AI tech has been enabling product visualisation on a range of different body types and skin colors for a couple of years already—sizes range from XXS to 4XL and a wide spectrum of skin tones are guided by the Monk Skin Tone Scale—this personalisation aspect is a first for the the tech powerhouse.
Although it comes in the wake of releases from other AI powered start-ups, Google dubs its own proprietary version “the first of its kind working at this scale” in that it allows shoppers “to try on billions of items of clothing from our Shopping Graph.”
According to the tech giant, “it’s powered by a new custom image generation model for fashion, which understands the human body and nuances of clothing — like how different materials fold, stretch and drape on different bodies.”
Virtual try-on app Doji recently announced a $14 million investment
Doji virtual try-on solution
Most hyped of Google’s more boutique competitors is Doji which this month announced a $14 million raise led by Thrive Capital with continued participation from 776. Doji, which was founded in 2024 by Apple, Meta, Shopify and Google DeepMind alumni Dorian Dargan and Jim Winkens, leverages its own proprietary diffusion models (generative algorithms that add ‘noise’ to a dataset to learn how to remove it leaving behind high quality images.)
It creates user avatars from six selfies snapped at different angles plus two full-body images on which it generates different suggested looks from brands including The Row, Sacai and Miu Miu while also allowing users to choose top and bottom pairings from listed items or uploaded urls. It directs users to partner sites for purchase—presumably working on an affiliate system.
Doji has recently launched on the App Store following a few months in private beta, however it is still invitation only as it continues to be finessed by the development team.
The downside is that avatar generation still stands at around 30 minutes but having tried it in its various incarnations, it offers arguably the most fidelity in terms of its photorealistic renderings when compared to competitors. Creation of further looks has speeded up considerably over the last few weeks.
Virtual fit solutions
The previous wave of virtual try-on tech which included studios like Bods focused more on virtual fit technology with the objective of decreasing returns to save brands money and minimise environmental impact. Via computer vision and machine learning it generated body avatars from a combination of photographs and measurements.
According to a story by Tech Crunch, Doji plans to address the issue of fit further down the line.
Wanna’s bag capacity widget
Wanna adds bag capacity widget to its virtual try-on suite
Elsewhere, Wanna, a provider of virtual try-on solutions in AR—most notably for jewelry and accessories—and the fashion arm of Perfect Corp, industry leader in VTO for beauty, has just released its animated Bag Capacity Widget, a new innovation for online bag shopping within its product suite.
It enables shoppers to assess storage capacity by virtually placing everyday items inside a bag, viewing capacity more accurately via a transparency mode that also reveals how many items would fit inside.
As owners of larger iPhones will doubtless attest, the studio notes that size mismatches account for around 24% of returns in the bag category.
Virtual try-on solutions are certainly a category to watch both as tools for engagement and those leading to more tangible return on investment.
