Why are artists so concerned about the Lensa app?
So, what exactly is this Lensa app, and why are creative types so concerned about it?
If you've been on social media at all lately, you may have seen some of your friends post digitally animated portraits of themselves that look like they came from another planet.
Even well-known people, like Chance the Rapper and Megan Fox, have joined in.
The portraits look like they were made by digital artists, and in a way, they were. But no one is getting paid or even getting credit for making them.
This is because these pieces were not made by people. Someone used a machine to make them.
With the Lensa AI app by Prisma Labs, your selfies can be turned into customised portraits using artificial intelligence. This lets users be whoever they want to be.
But even though it's a huge hit on social media, digital artists don't like the app because they say the works it makes are based on stolen art.
Not up to date? What you need to know is listed below.
How do people use Lensa?
Lensa AI makes it easy and cheap to make these digital portraits, two things that have no doubt contributed to the app's meteoric rise in recent weeks.
Statista says that more than 5 million people have downloaded Lensa AI since November, when the "magic avatar" feature was added.
Usually, a subscription costs about 100 euros per year, but users can get a 7-day free trial. Everyone has to pay an extra 4 euros to use the "magic avatar" feature, which makes 50 fantastic portraits.
Users have to post 10–20 selfies to the app before they can see their portraits.
Lensa tells them to take pictures from different angles and with different expressions on their faces to get the best results. It also tells them that the final images may have "inaccuracies or defects."
Some of these problems show up as extra arms and legs or heads that turn in strange ways.
After the app has had about 10 minutes to run the user's photos through its AI model, it spits out different avatars. Then, users can save them to their devices and share them on any social media platform they want.
In a long FAQ, Prisma Labs said that users' photos are deleted from its servers as soon as avatars are made. This is done to protect users' information.
The avatars are kept for "as long as it takes to provide our users with the service," according to Prisma Labs. Each avatar belongs to the user who made it.
As millions of people around the world fell in love with their magical avatars, online artist communities started to feel worried.
Not only did these AI-made portraits take away commissions from digital artists, but some of those artists' works were used to train the AI model that made them, often without their permission.
It's true that Lensa trains its AI with a copy of Stable Diffusion, an open-source neural network model. The model uses billions of images from all over the internet, which are put together in a set called LAION-5B.
If you've been on social media at all lately, you may have seen some of your friends post digitally animated portraits of themselves that look like they came from another planet.
Even well-known people, like Chance the Rapper and Megan Fox, have joined in.
The portraits look like they were made by digital artists, and in a way, they were. But no one is getting paid or even getting credit for making them. This is because these pieces were not made by people. Someone used a machine to make them.
With the Lensa AI app by Prisma Labs, your selfies can be turned into customised portraits using artificial intelligence. This lets users be whoever they want to be. But even though it's a huge hit on social media, digital artists don't like the app because they say the works it makes are based on stolen art.
Not up to date? What you need to know is listed below.
How do people use Lensa?
Lensa AI makes it easy and cheap to make these digital portraits, two things that have no doubt contributed to the app's meteoric rise in recent weeks.
Statista says that more than 5 million people have downloaded Lensa AI since November, when the "magic avatar" feature was added.
Usually, a subscription costs about 100 euros per year, but users can get a 7-day free trial. Everyone has to pay an extra 4 euros to use the "magic avatar" feature, which makes 50 fantastic portraits.
Users have to post 10–20 selfies to the app before they can see their portraits. Lensa tells them to take pictures from different angles and with different expressions on their faces to get the best results. It also tells them that the final images may have "inaccuracies or defects."
Some of these problems show up as extra arms and legs or heads that turn in strange ways.
After the app has had about 10 minutes to run the user's photos through its AI model, it spits out different avatars. Then, users can save them to their devices and share them on any social media platform they want.
In a long FAQ, Prisma Labs said that users' photos are deleted from its servers as soon as avatars are made. This is done to protect users' information.
The avatars are kept for "as long as it takes to provide our users with the service," according to Prisma Labs. Each avatar belongs to the user who made it.
Why should artists be afraid of lenses?
As millions of people around the world fell in love with their magical avatars, online artist communities started to feel worried.
Not only did these AI-made portraits take away commissions from digital artists, but some of those artists' works were used to train the AI model that made them, often without their permission.
It's true that Lensa trains its AI with a copy of Stable Diffusion, an open-source neural network model. The model uses billions of images from all over the internet, which are put together in a set called LAION-5B.
Stable Diffusion then uses these images to learn how to make new works that, according to Lensa, "aren't copies of any particular artist's work." The copyright law isn't clear about these datasets. On its website, LAION says that the datasets don't break copyright laws because they only list the URLs of images. Instead, they act as indices to the internet.
The CEO of Stability AI, the company that made Stable Diffusion, also defended the AI model and the way it learns from machines. But even if AI art can get past legal problems, it's not clear what the right thing to do is when it comes to these image generators.
Artists talked about how they can regain control of their work in the age of machine learning after Lensa AI's magic avatars went viral.
One of the resources available to them is the search engine haveibeentrained.com.Mat Dryhurst and Holly Herndon, two artists based in Berlin, made it.
It lets artists check to see if their work is in the LAION datasets used to train AI, and if it is, they can choose not to have it there if they want to.
Dryhurst told Euronews Culture in an email that Stability AI and LAION have agreed to honour any request from an artist to opt out.
The European Union's Data Protection Act (GDPR) also lets EU citizens contact LAION directly to ask that their name or a picture of them be taken out of the dataset.
But Dryhurst said that more needs to be said about consent and working together in the art world than just the technical way to opt out.
According to Dryhurst, the recent uproar over artificial intelligence-created art is most likely the result of people's growing resentment over how art has lost value in society over the past century.