Arek Dvornechuck
Branding Expert
New update: Adobe announced its Firefly Image 2 Model last October 10, 2023.
In this article, I share my thoughts and experiences with Adobe Firefly, an AI-powered tool from Adobe and I recently had the chance to try out its newest update—Firefly Image Model 2.
In this post, I'll go over this new AI generator: Its features, highlighting the new updates, and how to get access to it, then I will do a hands-on beta test.
Since this article focuses on the new updates, you may also want to check out my new video trying out the latest Firefly Image Mode 2 on my YouTube channel.
Before we start exploring the new features of this tool, let me quickly explain what it is and how it works.
Adobe Firefly is for now only available in the beta version—it’s an Adobe AI-powered text-to-image generator.
It's similar to Midjourney or Dall-e, which basically allows you to generate images from text prompts (text-to-image).
However, it will eventually transform into a family of creative generative AI models that will be available in all Adobe products.
What sets Adobe Firefly apart from other AI text-to-image generators is that it's going to be integrated into Adobe's apps, like Photoshop and Illustrator, making it incredibly convenient for creatives to use.
Just imagine how much time you could save if you could almost instantly render the ideas from your mind just by using these AI tools and without having to manually start every single design from scratch.
Firefly is a tool preparing to incorporate AI into popular Adobe apps, allowing users and you to effortlessly explore and refine any idea fast and easily.
This new tool allows users to generate all kinds of art, including illustrations, digital art, and paintings (just like Midjourney or Dall-E).
From its only two available tools before—Text to Image and Text Effects, Adobe Firefly now already has additional two extra tools, which are Generative Fill and Generative Recolor.
To give you an overview of these tools, Text to Image allows you to create unique images from a text prompt and apply style presets.
Text Effects allow you to create designs inside characters and words.
Generative Fill gives you the ability to remove objects or paint new ones from text descriptions by just using a brush.
Lastly, the Generative Recolor allows you to generate color variations of your vector artwork from a detailed description.
Besides that, there is a whole bunch of other new tools coming soon that will allow you to turn your sketches into a fully editable vector (an awesome feature for logo designers!).
Additionally, there are other new features—for example, for those who work on videos, which allow you to change the weather, mood, and atmosphere, and adjust colors and other settings of your videos.
Moreover, there are several other explorations in the pipeline as shown below, that we are excited about!
You can also check out the full list of tools that are coming soon on Firefly’s website (under the “About” page).
Unlike before, you can now quickly access Adobe Firefly by simply going to firefly.adobe.com and signing in with your Adobe ID or email address.
Some of you might ask about Adobe Firefly's native app—It is not available.
However, Firefly is accessible as a standalone web application that can be easily accessed on mobile browsers like Chrome, Safari, and Firefox—For more information, see system requirements.
As suggested in their access guide, It's important to use the latest version of your chosen browser to ensure a smooth experience.
Just like in the previous model, once you've gained access to Adobe Firefly, you can just start generating your art by inputting your text prompt, but now you will be able to use new features.
Adobe Firefly's latest Image 2 Model promises the ability to craft lifelike imagery.
It is an enhancement of the original Firefly Imaging Model, exclusively available via the Firefly web platform.
Boasting a superior model architecture, advanced training techniques, and enhanced image creation capabilities, achieving your artistic goals has never been more streamlined.
I recently got to try this new model and its new features, as well as the results, are exceptional.
To give you a comprehensive review of the latest update, let me share with you my experience using it by also showing a comparison between the previous version and the new one.
I started off by using a simple prompt—"A portrait of a motorcyclist standing in Time Square"
Upon checking the results in model 2, it gave me very usable and great images that looked like stock photos.
Compared to the generated images using model 1, most of the photos have mistakes.
For example, the photo below shows as expected from the prompt but there are a few flaws.
If you try to look closer, the man's face looks awkward and it looks like a badly photoshopped image.
Results-wise, the new version is definitely much better.
But wait, there's more in this new update, so let's check them out one by one.
Still, inside the Text to Image tool, you can basically scroll down and see the new tab "Match," where you can use your own image or simply choose from the gallery provided as an inspiration.
Using this feature will give you results that are more specific to how you want your photos will appear.
As an example, I chose one image inspiration from the gallery provided, regenerated it, and gave me another set of photos with the touch of the inspiration I chose.
In my personal opinion, this feature is helpful for brand designers like me to create consistent images for our website, brands, and so on.
The next thing we will explore is the Photo Settings—After generating your prompt, all you have to do is scroll down and you will be able to see a new tab "Photo Settings".
In this tab, you will be able to adjust the Aperture, Shutter Speed, and Field of View of the generated images according to your preferences.
Aside from that, you can also toggle "Auto" for automatic setting adjustments.
Another new feature is the Prompts Suggestions, which I think is very helpful when you're not sure about what you want it to generate.
This feature will appear by simply typing words or keywords in the prompt bar and it will give you suggestions just like when you're searching on Google.
It will give you great ideas that are related or commonly linked to the keywords you typed.
The next feature is Link Sharing, this can be utilized by simply clicking one of the generated photos and tapping the "More options" icon at the top right corner, then there you can see the "Copy link" option.
This new feature will allow you to share the images with anyone and not just simply share, but they will also be able to explore the prompt you used and check the settings you created for that image.
The last feature is the Negative Prompts—This can be utilized by simply going to the "Advance Settings" tab.
There will appear a blank bar where you can type keywords you want to remove from the generated images.
I tried this by typing the prompt "taxis traveling over the intersections and a big city highway"
From the original photos it generated, I wanted to remove the clouds and the yellow lines.
So in the blank bar, I entered "clouds" and "yellow lines," which prompts it to remove these objects from the photo.
After entering the keywords, simply regenerate and It will provide you with another set of photos without the objects you asked it to remove.
Based on the results of trying out this feature, there are still clouds shown in the regenerated photos but the yellow lines were omitted.
All in all, for me, this great feature is still helpful, especially when you are already satisfied with the original images it generated and just want to make tiny changes.
Adobe Firefly is a web-based application and can only be accessed online—there is no downloadable version available.
The release date for the full version of Adobe Firefly has not been disclosed yet, as it is currently in the beta stage.
Adobe Firefly is free to use as it is in beta, and there is no information available on the pricing for the full version.
Yes, Adobe Firefly is designed to be intuitive and user-friendly, making it easy to use.
Adobe Firefly uses AI generative technology and can be used to create a range of art styles, including abstract, landscape, and portrait.
Overall, Adobe Firefly is an impressive tool with a lot of potential.
Its great new features make it more usable and when it comes to the quality of the images, I would say that they are close to Midjourney.
I like how easy it is to use because of the great UI and many tools available in the sidebar that are being applied as additional tags to your prompt.
Unlike with other AI art generators, you don’t have to rewrite your prompt every single time—just apply the new styles or features as tags and remove them anytime you want.
By the way, I also urge you to watch my first YouTube video review of the previous Adobe Firefly model on my channel, so you will be able to compare it with the new version without any hassle.
I'm excited to see how AI will get implemented into Adobe CC and how it will improve our workflow as graphic designers.
What do you think about the latest update? Share your insights in the comment section below!
