Google Launches Doppl: Visualize Outfits on Yourself

Google’s latest innovation in the realm of virtual fashion, Doppl, offers users a novel way to visualize outfits on a digital version of themselves. Announced on Thursday, Doppl is currently available in the United States for both iOS and Android devices. This experimental app allows you to upload a full-body photo of yourself and then use screenshots of various outfits found online to see how they would look when worn. The process is straightforward: after uploading your photo, you can choose any outfit you see on social media, thrift stores, or friends, and Doppl will generate a picture of your AI lookalike dressed in it. What sets Doppl apart is its ability to convert these static images into short, AI-generated videos, allowing you to see how the outfit moves and fits on you in a more realistic manner. To test the app, I uploaded a photo of myself dressed in a casual T-shirt, shorts, and mismatched socks. My first virtual try-on involved one of Google’s sample outfits—a white and blue striped shirt. While the shirt was accurately rendered, Doppl struggled to correctly place the suggested skinny jeans, instead replacing my shorts with red ones and wrapping fabric around my calves like leg warmers. This issue persisted when I tried another outfit featuring distressed jeans; the app extended my T-shirt to the length of the shorts while failing to render the pants properly. In a more peculiar instance, I attempted to try on an outfit consisting of a striped button-down shirt and long, striped shorts. Doppl shortened the shorts significantly and replaced my actual feet with AI-generated, albeit somewhat realistic, ones. Despite these glitches, the app managed to generate footwear for outfits that did not originally show shoes, suggesting an adaptive approach to creating complete looks. Marina Galperina, my colleague, noticed a similar pattern when using mirror selfies to virtually try on outfits. Both of our AI lookalikes were noticeably thinned, giving us bobblehead-like appearances. However, when Marina and I used full-body photos taken by others, the app maintained a more accurate representation of our physiques. This inconsistency highlights the current limitations and areas for improvement in the app’s AI algorithms. Google has implemented certain guardrails to prevent misuse of the app. For example, images of more revealing outfits, such as bikinis, cannot be uploaded. Additionally, attempts to generate images of public figures, like former President Donald Trump, were met with restrictions. These measures aim to address ethical concerns and protect user privacy, although they may limit the app’s versatility. Doppl is part of Google’s ongoing efforts to enhance its virtual try-on capabilities. Earlier versions of this technology allowed users to see clothes on a diverse range of model bodies, but with Doppl, Google is taking a step further by enabling personalized avatars. The goal is to make the virtual try-on experience more engaging and accessible, helping users explore their style in a fun and interactive way. Google emphasizes that as a Labs experiment, Doppl is in its early stages and may not always produce accurate results. This acknowledgment suggests that the company is open to feedback and continual refinement of the app. Doppl’s technology combines machine learning and computer vision to manipulate and overlay clothing onto user photos. Google’s AI must analyze the structure and texture of both the user’s body and the chosen outfit to render a convincing image. The app’s current limitations, such as difficulty with pants and inconsistent body representations, indicate that the technology is still evolving and needs further development to handle complex clothing items and diverse body shapes more effectively. Despite these challenges, the potential applications of Doppl are significant. For e-commerce, it could provide a powerful tool for shoppers to make more informed purchasing decisions, reducing the likelihood of returns due to poor fit or appearance. In the fashion industry, designers might use the app to visualize their creations on a wider audience, facilitating design improvements and market research. Moreover, the app could offer a creative platform for individuals to experiment with different styles without the physical constraints of changing clothes multiple times. Industry insiders have mixed reactions to the launch of Doppl. While some see it as a promising step forward in personalized AI-driven fashion experiences, others are cautious about its current limitations and potential for misuse. The app’s ability to generate realistic movements in the AI-generated videos is particularly noteworthy, as it simulates the dynamics of clothing in a way that static images cannot. Google, known for its technological prowess and innovation in AI, has a track record of pushing boundaries with experimental projects. The company’s focus on ethical considerations and user privacy within Doppl aligns with its broader mission to develop responsible AI technologies. As Doppl continues to evolve, it will be interesting to see how it addresses current issues and expands its availability to other regions, potentially reshaping the way we interact with fashion and styling in the digital age. In summary, Doppl represents a significant advancement in virtual try-on technology, offering a unique blend of personalization and interactivity. While it faces some technical challenges, the potential benefits for both consumers and the fashion industry make it a fascinating experiment. Google’s commitment to ethical development and user feedback suggests that Doppl could become a valuable tool for style exploration and online shopping in the future.