Discussing generative AI within the art industry comes with a required analytical debate on ethics—and chief among these discussions is also concern over whether, one day, the technology could replace the artist itself. But, software company Skylum says that its new generative AI features inside Luminar Neo are designed to serve photographers, not replace them. Recently, Skylum announced that AI-focused editing software Luminar Neo will soon gain generative, erase, expand, and swap tools to its already broad line of AI-assisted tools.
The new tools, GenErase, GenExpand, and GenSwap, are the first forms of generative AI to join Luminar Neo’s other AI-based tools. All three tools are built using the Stable Diffusion model, which is the same model used in Adobe Photoshop’s generative tools. The three tools will roll out individually, one month at a time, through the remainder of 2023. GenErase launches first on Oct. 26.
Luminar Neo’s newest generative AI tools
While Luminar Neo’s existing tools are heavily based on artificial intelligence, the three new tools are the first tools that are considered generative AI.
- GenErase: Like a basic eraser, GenErase removes sections of an image. But, unlike the standard tool, GenErase fills in the missing area using the surrounding content. Content-aware fill existed before generative AI became a buzzword, but the AI analyzes the image to determine which parts of the image to sample from. And, if it’s anything like Photoshop’s similar eraser tool, it can also fill in with elements that aren’t a 100 percent match to the existing image.
- GenExpand: This tool essentially uncrops a photograph, allowing photographers to choose a different aspect ratio or create more negative space on the edges of the image. The AI generates new expanded borders for the image based on what already exists in the image.
- GenSwap: The final tool in the fall launch fits more in line with what most people think of when they hear generative AI. GenSwap can add an object that wasn’t there or swap existing objects out for ‘better’ ones.
Luminar Neo’s new tools are an expected launch for a software that’s so heavily AI-focused
The whole idea behind Luminar Neo is to streamline the editing process using AI. The software is designed to either save expert photographers time or help the less technically savvy jump right into advanced image editing. Because Neo is so AI-focused, I would be more surprised if the program didn’t eventually adopt some generative-based tools. The erase and expand tools were already listed on the software’s website as coming soon, which also shows a scene swap and water enhancer tool yet to come. The program already adds dramatic lighting to an otherwise flat image, adds sun rays, swaps skies, and corrects skin blemishes. Is adding objects that don’t already exist really much of a stretch beyond adding lighting that didn’t exist?
The new tools also aren’t terribly surprising because Adobe Photoshop has already launched similar tools. If Luminar Neo wants to be known as the AI-based photo editor, then it needs to remain competitive with the photo editor that’s so well known its name is also a verb. Photoshop’s recent updates add an AI-based eraser tool and edge expansion, as well as the option to type in text and have it spit out a generated object.
Where is the ethical line in generative AI photo editing?
In the press release announcing the latest features, Skylum describes Luminar Neo as software that uses AI “in a manner that serves them, not replaces them.” I believe that way of thinking is key to determining whether or not using AI in an image is ethical (alongside factors like how the image is used and labeled).
I’ve already used Photoshop’s generative AI eraser several times. The tool has saved me hours of stress editing, including removing the edge of a bra showing due to ill-fitting clothing and even editing out a strand of long grass that bisected a face. If the AI is doing what photographers have already been doing for years, like removing power lines and street signs for fewer distractions in the background, then there’s really no ethical question. (Excluding, of course, mislabeling the image as unedited or using within stricter genres like photojournalism). Why not have more time to do what you love when the end result is the same? That’s the exact reason why I bought the camera that I did—time saved editing—and spending less time behind the computer can only serve photographers. GenErase and even GenExpand, much of the time, fall within those ethical boundaries.
GenSwap may toe the line—or even cross it at times—because it adds something that wasn’t already there. I’ve turned photobombers into yetis for Facebook laughs, but otherwise, I haven’t ventured into adding objects to my photographs. But is adding objects to a photograph going to compete with artists? Considering that photobombing yeti turned into an actual big foot rather than Big Foot with generative AI, the answer, at least right now, is no. Of course, the other half of the equation is just how well the software works — a question that cannot be answered until each tool rolls out.
The new generative tools will be added for existing subscriptions, while users who purchased perpetual licenses can add the new features for $69 before Oct. 28, which will also include any new features launched before Aug. 16, 2024.