Walmart's Virtual Try-On - Tech Insiders News, September 2022
A curated collection of «signal over noise» links about interactive technologies, AI, web3, sense-making, entrepreneurship, and psychology.
Cover image generated by Midjourney, prompt: «Whispers of fashion in a carbon neutral virtual world» (Inspired by the links below.) Full image at the bottom of the article.
- Walmart is rolling out a new feature in their app that allows users to virtually try on clothes by taking a guided selfie.
Why it matters: I don’t have data on hand, but I remember that fashion is one of the industries with the most returns after purchasing through e-commerce. This is a good step towards increasing consumer trust!
I wonder if they also give a size estimation? Because this is the tricky part for any virtual try-on. It works fine for testing the style, but the size is another complex issue! - OpenAI releases Whisper, their latest open-source speech-to-text program.
Why it matters: the developer ecosystem is currently on fire and jumps on immediately as soon as a new «AI» is released. It seems that Whisper is exceptionally efficient and might revolutionize transcription and traduction. - Nvidia’s GTC (GPU Technology Conference) was held last week, with a lot of exciting news. I’m picking two of them. Omniverse Cloud is a suite of building blocks toward Virtual Worlds* as a service. And Jensen Huang, Nvidia’s CEO, explained how the future virtual worlds would be auto-generated/populated with the help of AI. (Link to the paper here)
Why it matters: Nvidia depicts - and builds - a future where the creation of the virtual worlds will be accessible to anyone. I can describe the world using my voice, which will be generated automatically on the cloud!
*I could have said Metaverse, but let’s remove the buzzwords, shall we?
And as a thought experiment, I asked Stable Diffusion to generate «A top view of a video game map, with forests, beaches, rivers, and mountains.» That’s a start! - There were a lot of fashion weeks this month. Fashion and luxury is the industry that takes the Metaverse topic most seriously, with NFTs, collaborations, and presence in many Virtual Worlds. This week I am selecting the Virtual Fashion Show done by the sportswear Anta, in partnership with Baidu and the Style3D studio.
Why it matters: A brand using a Virtual Word is modern (to attract GenZ), cost-effective, and environment-friendly. I’m not sure about the latest one; all these servers and GPUs and consuming a lot of energy. But it’s still better than doing a fashion show in the desert and flying in models and crew, right? Oh wait, but it seems it was carbon neutral… Uhhh… Everything is complex, right? - Related, both on societal and sense-making levels. So, one technique to create AI models is to take a lot of data, for example, pictures, and then classify it manually. An operator will draw an area around the objects of the image: this is a car, bike, etc. I found two articles about that: one that depicts the poor conditions of such operators in Kenya and another that praises the jobs created in India. Two different views on similar jobs. These jobs are repetitive and low paying, but can one consider that these workers are better off in front of a computer 8 hours a day than if they didn’t have any job? Who are we to judge that? And are these jobs doomed anyway when data will be trained in virtual worlds?