Adobe’s Firefly artificial intelligence image generator is added to the software

Iceland’s scary part: from waterfalls to foregrounds, and back again: A picture of Skgafoss with Generative Fill

If you would like to visit a place where danger is abundant, and where lunch is $60, I encourage you to visit Iceland. It is a beautiful and brutal landscape whose natural wonders (and terrors) make for incredible photographs. I took a bunch on a recent trip to the country. But it is possible, as I’ve unfortunately discovered, to make them terrifying in entirely new ways.

A new generative artificial intelligence tool will allow users to quickly add or remove objects with text prompt. The feature is called Generative fill and is one of the first Creative Cloud applications to use Adobe’s artificial intelligence image generator Firefly. Generative Fill is in its alpha stage, but Adobe says a full release will come later this year.

After messing around with it for just a few hours, it already looks like an impressive upgrade to the existing Content-Aware Fill tool in Photoshop. There are people in the photo of Skgafoss. The original photo has a lot of people in front of the falls, so my photo was naturally made to look that way.

The Generative Fill picture is better than the Content-Aware picture. I had to type in a few words, wait for a few seconds, and make an error-free selection. It is kind of scary.

If you really want to go for it, you can do it. And oh, the places you will go with Generative Fill. From a beautiful waterfall to a dead glacier is a journey we should take. Here’s what the valley carved out by the glacier actually looked like, and here’s what I got when I asked Generative Fill to make it even more ominous by adding a thunderstorm.

Generative fill goes hard on the waterfall. It’s kind of phony looking, but plausible, I guess. Now, what if I ask for ponies in the foreground instead?

Cool, the world’s bleakest lemonade stand. Love it. How about we lighten up with something that couldn’t possibly conjure up horror? Is there a rainbow in the sky?

Okay, that’s an F for that rainbow and just a big “Nope” to the rest. You okay, Firefly? Take me back to safety and just try to remove the people from this photo at the beautiful-yet-deadly black sand beach.

Towards a Safe and Legally Secure Use of Generative Fill for Outpainting and Generating New Objects in Iceland

It is okay. I guess. The area where the rocks meet the water looks fake and stretched, but honestly, the whole country looks too beautiful to be real. And at least the phallic image in this photo wasn’t made by AI. Seriously, it’s just out there like that. You never know what you’re going to see next in Iceland.

Generative Fill is a tool that can be used within a file in a common image editing program. If you use it to expand the borders of an image (also known as outpainting) or generate new objects, it’ll provide you with three options to choose from. When used for outpainting, users can leave the prompt blank and the system will try to expand the image on its own, but it works better if you give it some direction. It is similar to the content-aware fill feature in that it offers more control for the user.

Such feats won’t be a huge surprise for creators familiar with AI image generation tools, but, as ever, it’s the integration of this technology into mainstream apps like Photoshop that bring them to a much wider audience.

The training data is an important part of Firefly. Adobe says that the model is only trained on content that it has the right to use, such as Adobe Stock images, openly licensed content, and content without any copyright restrictions. This means if you’re using Generative Fill feature, you should be able to use it safely, as it will be more transparent about their training data. This will likely be a consolation to creatives and agencies who have been wary about using AI tools for fear of potential legal repercussions.

Generative Fill also supports Content Credentials, a “nutrition label” system that attaches attribution data to images before sharing them online, informing viewers if the content was created or edited using AI. You can inspect an image by using verify.contentauthenticity.org where you will find an overview of information.