The Lensa photo and video editing app has shot into social media prominence in recent weeks, after adding a feature that allows you to generate stunning digital portraits of yourself in contemporary art styles. It does that for only a small fee and the trouble of uploading 10 to twenty different photographs of yourself.
2022 has been the yr text-to-media AI technology left the labs and commenced colonising our visual culture, and Lensa would be the slickest business application of that technology thus far.
It has lit a hearth amongst social media influencers trying to stand out – and a distinct kind of fireside among the many art community. Australian artist Kim Leutwyler told the Guardian she recognised the sorts of particular artists – including her own style – in Lensa’s portraits.
Since Midjourney, OpenAI’s Dall-E and the CompVis group’s Stable Diffusion burst onto the scene earlier this yr, the benefit with which individual artists’ styles could be emulated has sounded warning bells. Artists feel their mental property – and maybe a little bit of their soul – has been compromised. But has it?
Well, not so far as existing copyright law sees it.
If it’s not direct theft, what’s it?
Text-to-media AI is inherently very complicated, nevertheless it is feasible for us non-computer-scientists to grasp conceptually.
To really grasp the positives and negatives of Lensa, it’s price taking a few steps back to grasp how artists’ individual styles can find their way into, and out of, the black boxes that power systems like Lensa.
Lensa is actually a streamlined and customised front-end for the freely available Stable Diffusion deep learning model. It’s so named since it uses a system called latent diffusion to power its creative output.
The word “latent” is essential here. In data science a latent variable is a top quality that may’t be measured directly, but could be be inferred from things that could be measured.
When Stable Diffusion was being built, machine-learning algorithms were fed a lot of image-text pairs, they usually taught themselves billions of other ways these images and captions might be connected.
This formed a posh knowledge base, none of which is directly intelligible to humans. We might see “modernism” or “thick ink” in its outputs, but Stable Diffusion sees a universe of numbers and connections. And all of this derives from complex mathematics involving the numbers generated from the unique image-text pairs.
Because the system ingested each descriptions and image data, it lets us plot a course through the big sea of possible outputs by typing in meaningful prompts.
Take the image below for example. The text prompt included the terms “digital art” and “artstation” – a site that’s home to many contemporary digital artists. During its training, Stable Diffusion learnt to associate these words with certain qualities it identified in the varied artworks it was trained on. The result’s a picture that might fit well on ArtStation.
Stable Diffusion
What makes Lensa stand out?
So if Stable Diffusion is a text-to-image system where we navigate through different possibilities, then Lensa seems quite different because it takes in images, not words. That’s because one in all Lensa’s biggest innovations is streamlining the means of textual inversion.
Lensa takes user-supplied photos and injects them into Stable Diffusion’s existing knowledge base, teaching the system “capture” the user’s features so it could possibly then stylise them. While this be done within the regular Stable Diffusion, it’s removed from a streamlined process.
Although you may’t push the pictures on Lensa in any particular desired direction, the trade-off is a wide range of options which are almost at all times impressive. These images borrow ideas from other artists’ work, but don’t contain any actual snippets of their work.
The Australian Arts Law Centre makes it clear that while individual artworks are subject to copyright, the stylistic elements and concepts behind them should not. Similarly, the Dave Grossman Designs Inc. v Bortin case within the US established that copyright law doesn’t apply to an art style.
What in regards to the artists?
Nonetheless, the incontrovertible fact that art styles and techniques at the moment are transferable in this fashion is immensely disruptive and intensely upsetting for artists. As technologies like Lensa becomes more mainstream and artists feel increasingly ripped-off, there could also be pressure for laws to adapt to it.
For artists who work on small-scale jobs, resembling creating digital illustrations for influencers or other web enterprises, the longer term looks difficult.
However, while it is straightforward to make an artwork that appears good using AI, it’s still difficult to create a really specific work, with a selected subject and context. So no matter how apps like Lensa shake up the way in which art is made, the personality of the artist stays a vital context for his or her work.
It could also be that artists themselves might want to borrow a page from the influencer’s handbook and invest more effort in publicising themselves.
It’s early days, and it’s going to be a tumultuous decade for producers and consumers of art. But one thing is of course: the genie is out of the bottle.
Stable Diffusion
This article was originally published at theconversation.com
