Hands On with Adobe Firefly: Generative AI for Creators Adobe take on the Generative AI trend

Adobe recently announced a new beta of its Firefly generative AI service designed specifically with content creators in mind. Firefly can create images from text, but adds some interesting features to the mix as well as create AI-based text effects and more.

In the video above Jesús Ramirez (from PTC) does an early comparison between the latest version of Midjourney and the new beta of Adobe Firefly with some interesting insights.

Images from text (with a spin)

Generative AI which can create text, images, and even (still relatively basic) video from a user textual input is gaining more and more momentum with new services pooping almost every week and an extremely fast paste of development never seen before since the invention of the personal computer (users are seeing measurable improvements in a matter of months and sometimes even weeks).

With all of this going on it was clear that Adobe, which had already been working on integrating AI into is different software packages can not be left behind. This week the company introduced its latest tool called Firefly.

This new generative AI service is still very much in beta (you need to have an Adobe ID and request access) but it is clear that from Adobe’s perspective, this is just the first step before integrating this new technology into its existing line of products (and maybe adding a few new ones).

So how is the new Firefly image from text is different than the latest version of Midjourney as an example? Probably the first thing that you will notice is that Firefly has a very different interface that allows you to more easily focus your ideas and create more specific results. This makes a lot of sense (actually we would like to see many more options in future versions of this interface), especially for a service that eventually is aimed at professional creatives who needs to work fast and produce very accurate results (you can think of Midjourney in this context as a more artistic tool).

As for the results, the Generative AI was trained on free Adobe stock images and the number of images that it has been trained on at the moment is much smaller than that of the Midjourney algorithm although is quite clear where the stronger and weaker aspects of each of these tools are.

At the moment it seems that Midjourney is still ahead but the margin is not as big as you might think which is surprising as Firefly is still very much in beta while Midjourney already had several public iterations.

Official Adobe Firefly video

YouTube player

Personalized AI-based Text Effects

Firefly has an interesting second feature that we are seeing from a generative AI in this specific form for the first time and that is text effects. This allows the user to write any text and add to it one of many (many) effects (anything from a tiger pattern to a cookie).

This is a very specific use case but it really fits the  Adobe creative ecosystem and can be a great addition to Photoshop and other design tools that Adobe have in the future.

AI-based brushes and more

Adobe already has another upcoming feature in Firefly called “Recolor vectors” which is said to create unique variations of your artwork from a detailed text description. This is certainly interesting but an even more interesting feature Adobe is working on is AI-based brushes.

Such a tool will enable users to easily create new brushes designed by AI for their specific workflow. We have no idea when this tool is coming but it can certainly be interesting.

From our own personal perspective what we would like to see is a type of content-aware AI that uses a sample from a specific part of an image and knows how to accurately fill a selected area of an image (Adobe has been making steps in this direction but in our experience, it’s not really 100% their yet).

Recolor vectors

More Firefly coverage

Since it was announced earlier this week several channels have published hands-on videos with Firefly – here are a few of them:

Terry White from Adobe with a first hands-on of the Adobe Firefly

YouTube player

Jesse Showalter on Firefly

YouTube player

DesignCourse on Firefly vs. Midjourney 

YouTube player

 photoshopCAFE look at Firefly

YouTube player
Iddo Genuth
Iddo Genuth is the founder and chief editor of LensVid.com. He has been a technology reporter working for international publications since the late 1990's and covering photography since 2009. Iddo is also a co-founder of a production company specializing in commercial food and product visual content.

Leave a Reply

Your email address will not be published. Required fields are marked *