“I think I have a pretty decent self-image, but I look at pictures and I’m like, ‘Why do I look so good?’ I said,” he said. a media presence separate from his day job. “I think it shaved off a lot of my rough edges.”
Lensa, a photo and video editing app from Prisma Labs, has been available since 2018, but its worldwide downloads have skyrocketed since the launch of its “magic avatars” feature in late November, according to analytics firm Sensor Tower. In the first five days of December, the app saw 4 million installs, compared to 2 million in November, topping the Apple and Google app stores. Sensor Towers reports that consumers spent $8.2 million on the app during that five-day period.
The app is subscription-based and costs $35.99 per year, with an additional $3-$12 fee for avatar packs. Upload eight to 10 photos of your face, with no one else in the shot, filling most of the frame, and Lens will use the photos to train a machine learning model. The model then creates images based on your face in various artistic styles, such as “anime” or “fairy princess”.
Some people were amazed at how flattering or accurate the portraits looked. Others shared distorted images with distorted facial features or limbs sticking out of their heads, Lensa warns during the upload process.
The trend has also raised concerns about the fairness of AI-generated images, the impact on professional artists and the risk of sexual exploitation. Here’s everything you need to know before downloading.
Lensa is owned by Sunnyvale, Calif.-based Prisma Labs, which also makes Prisma, an app that uses artificial intelligence to duplicate photos in a variety of artistic styles. Prisma Labs CEO Andrey Usoltsev and co-founder Alexei Moiseenkov previously worked at Russian tech giant Yandex, according to their LinkedIn profiles.
Like rival Facetune, Lensa comes with a collection of photo and video editing tools that do everything from replacing your cluttered living room with an artistic backdrop to removing bags under your eyes.
How does Lensa create AI avatars?
Lensa is based on a free-to-use machine learning model called Stable Diffusion, trained on billions of image and text combinations pulled from the web. When you upload your photos, the app sends them to the cloud and creates a personalized machine learning model for you. Then that model spits out new images in your likeness.
He used artificial intelligence to win an art competition. Was it cheating?
Will the pictures look like me?
It depends. Some dark-skinned users say they see more flaws and distortions in their avatars than their lighter-skinned friends, fueling long-standing concerns about fairness in AI portrayal. Asians and people wearing headscarves have also shared inaccuracies in AI portraits on Twitter.
Usoltsev did not address concerns about the app’s alleged tendency to translate results into English, referring The Washington Post to frequently asked questions posted on the Prisma Labs website.
Because of both AI engineering and the lack of representation of dark-skinned people in the training images, the models analyze and reproduce images of dark-skinned people worse, says Mutale Nkonde, founder of AI for People’s Algorithmic Justice. For example, in scenarios where facial recognition is used for law enforcement, it creates frightening opportunities for discrimination. The technology has already contributed to at least three wrongful arrests of black people.
Nkonde noted that there is potential to harm Lensa as well. From what she’s seen, the app’s results for women tend toward the “generic hot white girl,” she said.
“It can be very damaging to the self-esteem of black women and girls,” she said. “Black women look at it and go, ‘Huh.’ I love the picture. It doesn’t look like me. What’s going on with that?’”
Because Lens allows you to choose the gender of your avatar, including an option for non-binary, some trans people have noted the opportunity to see a gender-affirming version of themselves.
Trans peeps: If you’re into Lensa, take a bunch of old photos of you as a teenager and pre-transition, enter your true gender in the survey, and run them in the app.
You’ll get a bunch of images of your younger self: pic.twitter.com/5CBGRGkpfA
— June (@beloved_june) December 4, 2022
Should I be concerned about privacy?
It also says that it only uses the photos you provide to create an avatar, and once the process is complete, it deletes each set of photos along with a machine learning model trained on your photos.
Usoltsev said Prisma Labs does not use photos or personalized models to train its facial recognition network. He declined to say whether Prisma Labs keeps any data based on your photos, but said the company keeps “minimum.”
The real privacy issue with Lens comes from a different angle. A huge collection of images used to train an AI called LAION has been unintentionally removed from the internet, according to AI experts. This means that it includes images of people who do not consent. One artist even found photos from his personal medical records in the database. Visit HaveIBeenTrained.com to check if images related to you have been used to train an AI system. (This engine does not save your image searches.)
There is also the potential for exploitation and harassment. Users can upload photos of anyone, not just themselves, and the app’s female portraits are often nude or shown in sexual poses. This also happens with pictures of children, although Lensa says the app is only for people 13 and older.
“The Fixed Diffusion model was trained on unfiltered internet content. So it reflects the biases that people put into the images they produce,” Lensa said in an FAQ.
Artificial intelligence can now create any image in seconds, bringing wonder and danger
Why the backlash from digital artists?
Some creators have embraced the AI description with gusto. But as Lensa avatars took over social media feeds, many digital artists asked people to think twice before paying for the app. The artists say that Lensa’s “styles” are based on real art by real people and that these professionals are not compensated.
“No one really understands that a program that takes everyone’s art and then creates concept art actually affects our work,” said Jon Lam, story artist at video game company Riot Games.
Machine learning recreates patterns in images, not individual artworks, Lensa said in an FAQ.
But Lam said friends lost their jobs after employers used their creations to train AI models — the artists themselves are no longer needed in the eyes of companies, he said. In many cases, LAION infringed on the copyright of the images, and Prisma Labs said it was profiting from the artists’ life’s work without their consent. Some creators have even found it similar to the signatures of artists Internal images created in Lensa.
“Signature details are seen in styles that mimic paintings,” says the Lensa FAQ. “This subset of images often comes with signatures from the author of the artwork.”
If you want illustrations of your own that support traditional artists, find someone local or search sites like Etsy and order a portrait, Lam suggested.
“I see a really bad future if we don’t get this thing under control now,” he said. “I don’t want this to happen, not just for the artists, but for everyone affected.”