Meta brings image and text generation to Advantage+ to make creating ads easier

Meta is bringing full image and text generation to its Advantage+ offering, enabling advertisers to create compelling and creative ads rapidly. It’s another step by the company to automate certain parts of the ad creation process, which Meta believes will free up resources and improve ad performance. These new capabilities will be available globally later this year.

“For the first time, you can create full image variations inspired by your original ad creative and with text overlay capabilities,” John Hegeman, Meta’s head of monetization, said during a press conference. “Imagine you are promoting your coffee business by advertising a fragrant cup of coffee. Our generated AI will be able to create variations of your ad creative, including scenery that embodies a lush farm and also provide adjustments to the coffee cup to offer you more creative options to choose from.”

Besides generating an image through the use of a prompt a la Midjourney, Adobe Firefly or OpenAI’s DALL-E, advertisers can apply text on top of the image, choosing from one of a dozen popular font typefaces. There’s even an option to repurpose created images so the creative will suit different aspect ratios—something Meta rolled out in October with Instagram.

Advertisers will be able to utilize Meta’s AI to help craft impactful headlines and body copy for their advertisements. The company shared that it is also actively testing the ability to have the generated text reflect the advertiser’s brand voice and tone.

“We know that continually iterating on your ad creative is one of the most effective performance levers advertisers have,” Hageman remarked. “Marrying generative AI with our Advantage+ automation products enables us to drive maximum performance for businesses with greater scale, speed and accuracy…What we’re hearing from advertisers is that these generative AI tools are saving time and resources while increasing productivity.”

Meta revealed it will train these features on Llama 3, its next-generation large language model, which should provide better query results. “What you see is that when you replace a less advanced foundation model with a more advanced foundation model, do some fine-tuning on top, you’ll generally get higher quality results [and] better performance…I think it’s a pretty substantial difference,” Hegeman explained.

It’s not surprising to see AI being used to expedite the ad creation process, though designers, writers and marketers may be concerned they have less to do. However, as it stands today, Meta’s AI will generate images and text based on existing assets—it will not create things from scratch. The goal is to leverage what’s already there and expand upon in, producing variations to see which can generate the best return on investment.

Hegeman agreed that advertisers wouldn’t want to completely surrender the wheel to AI: “There are things that people are going to want to stay in close control of treating the first version of a specific campaign. And I think these tools are often most helpful for taking the labor out of some of the more grunt work of creating many different specific variations of something once you’ve got the basic idea.”

What about licensing, copyrights and all the other issues that come with using AI-generated media? Meta says it’s in “the process of working through some of the specifics” regarding how that plays out with ads and hopes to have more to share in the future.

Both the image and text generation capabilities, along with the other gen AI tools Meta affords advertisers, will be available in Advantage+ creative, the company’s AI tool repository for the campaign creation process.