Google’s Doppl App Blew Me Away – Must-See Features!
▼ Summary
– Doppl is a new AI-powered app by Google that lets users virtually try on outfits by generating images or videos of themselves wearing clothes from screenshots.
– The app struggles with accurately rendering pants when users upload photos of themselves wearing shorts, sometimes creating odd distortions like leg warmers or fake feet.
– Users need a full-body photo in bright light and a screenshot of the desired outfit to generate try-on images, which can then be animated with poses like waving or smiling.
– Doppl has guardrails to block revealing outfits or images of public figures, preventing misuse like generating explicit content or deepfakes.
– The app sometimes alters users’ body proportions in mirror selfies, making them appear thinner, though this issue doesn’t occur with photos taken by others.
Google’s experimental Doppl app offers a futuristic way to virtually try on clothes, though its AI-generated results sometimes miss the mark. The app creates surprisingly realistic animations of users wearing outfits sourced from anywhere online, blending personal photos with digital fashion in seconds. While the technology shows promise, early testing reveals some amusing glitches that highlight the challenges of AI-powered styling.
Getting started with Doppl couldn’t be simpler. Users just need two things: a clear full-body photo of themselves (no hats, good lighting) and a screenshot of any outfit found online. The app then works its magic, first generating a still image before offering the option to animate the result with various poses. During testing, the process took mere minutes to transform static images into dynamic clips showing the digital version of the tester waving, smiling, or striking other natural-looking poses.
The app’s most noticeable limitation appears when working with certain clothing items. Attempts to virtually try on pants over real-life shorts produced bizarre results, with the AI sometimes creating leg warmer-like fabric wraps instead of proper trousers. In one particularly odd instance, the system replaced mismatched socks with entirely fabricated feet, complete with convincing toenails, when the original outfit image didn’t include footwear. These unexpected “creative liberties” demonstrate both the technology’s potential and its current imperfections.
Doppl includes built-in safeguards against misuse. The app automatically blocks attempts to upload revealing outfits like swimwear or images of public figures, preventing potential abuse of the technology. However, testers discovered another curious behavior, when using mirror selfies as source images, the AI consistently generated thinner versions of the subjects, creating disproportionate bobblehead-like figures. This distortion didn’t occur with standard full-body photos taken by others.
Compared to Google’s existing virtual try-on features, Doppl represents a significant advancement. While previous tools only worked with clothing found in Google search results, this experimental app pulls from any online source and adds animated elements. The ability to visualize how random Pinterest finds or Instagram outfits might look in motion adds a new dimension to online shopping. With some refinement to its clothing recognition and body mapping algorithms, Doppl could revolutionize how people experiment with fashion digitally.
Currently available for both Android and iOS devices, Doppl gives anyone curious about AI-powered fashion a chance to test its capabilities firsthand. The app’s occasional quirks provide entertaining examples of how machine learning interprets human clothing choices, while its successful transformations hint at a future where virtual wardrobe experiments become seamless. As the technology develops, these digital dressing rooms may soon eliminate the guesswork from online shopping altogether.
(Source: The Verge)
