Adobe Firefly

About

NOTE: This page is a daughter page of: Artificial Intelligence

Adobe Firefly is Adobe's answer to products like Midjourney and it has a beautiful UI and capability. Firefly was unveiled March March 21, 2023, as a family of new creative generative AI. The Adobe Firefly model will empower customers of all experience levels to generate high-quality images and stunning text effects. Try it here: firefly.adobe.com.

Early teaser (preview)

Adobe Firefly Features

Some of the different firefly modes as of Oct 2013 - more coming soon.

Text to Image Mode

Style Reference Image

Firefly seems to have a particularly easy interface, and the option to choose or upload a "style reference image" is a huge drawcard! Brilliant. Sadly you can only upload one image right now, but it matches the style well.

An example of using a reference image to generate a new character but with a similar style (children's book style here).

Remove Background

When you pick an image, it has the options you'd expect (download, favorite, save to library, etc) plus some other good ones "Show Similar" (to generate variations) and my favorite being "Remove background" (which opens Adobe Express for some quick editing). Brilliant ideas!

Some of of the options for an image (not shown: on the download icon are options to: download, copy link, copy image, edit in Adobe Express, save to library).

The same egret in Adobe Express after clicking "Remove Background".

Creating Consistent Characters

I don't feel like this is a strength for Adobe Firefly. Maybe it will be someday, but here's a video that gives some strategies:

• Adobe Firefly AI Consistent Characters

• Notice:
  • Her prompt is: Chibi Bunny with a blue backpack, Detailed Character Design Sheet, in 8 poses.
  • She also adds lots of style options.
  • She has a similar video "Adobe Firefly AI Consistent Character Pose" explaining how you could change the background.

Video of bunny creation.

Another tricky you might use is Generative Fill, where you try to just change one aspect of an image at a time.

Generative Fill Mode

Generative fill is an incredible, incredible free Firefly tool to take any area of your image you desire, and tweak it or replace it with something completely different. You can change the background, or a person's clothes, or add a dinosaur into the room somewhere... anything. Make tweaks then hit "Keep" to make the next tweak. Very fun! I guess as the final stage you'd probably want to go to Photoshop to touch up any edges that look weird or paste in other characters. Better yet, if you pay for Adobe, they added a Generative Fill option directly into Photoshop (see demo video), so you can easily change the crop of your image to zoom out and expand the scene to something larger.

Generative fill for "fancy blue suit holding a parrot" (I tweaked it since this first screenshot).

Generative fill result, after changing clothes.

Trying to change the characters pose by adding a drink into his hand. I then went on to change the eye and open the beak... and changing the background is a piece of cake.

Text Effects Mode

Super cool, super intuitive! Only a finite number of fonts, but you can type pretty well anything... and I'm sure they'll add a feature to generate any shape mask soon enough - if not already.

Text effects, with loose fitting. "Australian children's book with birds, leaves and gumnuts".

Videos

• Adobe Firefly vs. Midjourney v5 - Same Prompt Comparison - Some amazing examples - quite comparable.... but Firefly promises to do a bunch of stuff like decorative text that Midjourney cannot yet handle. They both have pros and cons.

See Also

• Midjourney - Biggest competitor in text-to-image so far as I can tell.

Links

• Adobe Firefly - official website.