Adobe Firefly Boards Explained

Firefly Boards Is Adobe’s AI Workspace, and It Runs in a Browser Tab

Most AI image tools give you one prompt box and one result at a time. You type, you wait, you look, you type again. It’s a slow way to explore an idea, and it’s the reason so much AI work stops at the first output that isn’t obviously wrong.

Firefly Boards is a different shape: an infinite canvas in your browser where images, prompts and generations all live side by side, and where you can fire off several generations at once without waiting for any of them. It also gets Adobe’s newest models — including third-party ones like Gemini 2.5 and the Veo video models — before they reach Photoshop. This walkthrough covers merging two photographic styles and then animating the result.

What Makes a Canvas Different From a Prompt Box

Generations Run in Parallel

Hit Generate three times with different settings and all three run at once, landing on the canvas as they finish. Comparing options side by side is a fundamentally better way to choose than judging one image in isolation.

Your Source Images Stay in View

The subject photo and the style reference sit on the same canvas as every generation. You can see at a glance how far a result has drifted from what you asked for.

You Get the New Models First

Boards is where Adobe ships early access. If a model is available anywhere in the Adobe ecosystem, it usually appears here before it appears in a desktop application.

Set Up the Board

1. Open Firefly Boards and Create a New Board

   Everything runs in the browser — there’s nothing to install, and the Firefly Boards canvas saves to your Adobe account rather than a local file.

2. Drag Both Images Onto the Canvas

   Your subject photo and your style reference. The style reference is doing more work than people expect: it’s supplying the color palette, the light quality and the grade, so pick one with a look you can actually name.

3. Arrange Them With the V Key

   V is the select-and-move tool, as it is in the rest of the Adobe suite. Put the two images somewhere you can see both while you work — the canvas is infinite, and generations will fill it fast.

Merge the Two Styles

4. Open Generate Image and Pick a Model

   The model list includes Adobe’s own Firefly models alongside third-party ones. Gemini 2.5 Nano Banana is a good default here — it’s fast, and it handles style transfer between two supplied images well.

5. Load Both Images With the Eyedropper

   Use the eyedropper to pull your subject and your style reference into the prompt box. This is the step that distinguishes a style merge from a text-prompted generation: the model is working from your actual images, not from a description of them.

6. Write the Edit and Hit Enter

   In Describe the edit, something as plain as “Merge the style of these two photographs” is enough. The images are carrying the information; the text just tells the model what to do with them.

7. Generate Several Times Without Waiting

   Click Generate again immediately. Runs queue in parallel and appear on the canvas as they complete, so three or four variations cost roughly the same wall-clock time as one.

Refine the One You Want to Keep

8. Click Load on Your Favourite

   This pulls that image’s original prompt and settings back into the generation panel — model, prompt, everything — so you’re editing a known starting point rather than trying to reproduce it.

9. Change the Aspect Ratio and Regenerate

   Switch horizontal to portrait for social, or the reverse. Regenerating at a new ratio gives you a properly composed vertical rather than a crop of a horizontal image, which is the usual reason reformatted AI work looks wrong.

10. Click Place on Canvas

    The refined version joins the Firefly Boards canvas alongside everything else, so you keep the earlier attempts for comparison.

Turn the Still Into Video

11. Switch to Generate Video and Choose a Model

    Video 3.1 handles this well. Video generation lives on the same canvas, so there’s no export step between the still and the clip.

12. Set Your Image as the First Frame

    This is what keeps the video anchored to the picture you spent the last twenty minutes getting right, rather than generating something new that merely resembles it.

13. Describe the Motion, Not the Scene

    Something like “slowly shake head left to right while zooming in” works because it describes movement. The scene is already established by the first frame — re-describing it just gives the model room to change things you liked.

14. Download the Clip With Its Audio

    The result comes out at high resolution with natively generated audio included. Press Cmd/Ctrl + S to save the Firefly Boards canvas before you export, so the whole exploration is still there next time.

What to Watch Out For

A Style Reference With No Clear Style

If you can’t describe what makes your reference image look the way it does, the model can’t extract it either. Pick references with an obvious, consistent grade.

Generating One at a Time

The parallel generation is the main advantage this tool has over a prompt box. Using it serially throws that away.

Cropping Instead of Regenerating for a New Ratio

Load the prompt, change the ratio, generate again. It takes seconds and gives you a real composition instead of a truncated one.

Over-Describing the Video Prompt

Describe motion only. Every extra detail about the scene is an invitation for the model to reinterpret a frame you’d already settled.

Where This Fits in a Real Workflow

Boards is best understood as a place to explore rather than a place to finish. Generating twelve directions in the time a desktop tool would take for one is genuinely useful at the front of a project, when the question is still what the thing should look like.

What comes out is a strong starting point, not a final file. The retouching, the precise color work and the compositing decisions still belong in Photoshop — and knowing how to do them by hand is what lets you tell a good generation from a merely impressive one.

Open a Board and Try It

Create a board, drag in a subject and a style reference, eyedropper both into Generate Image, ask it to merge the styles, fire off several generations at once, reload the best one at a new aspect ratio, then send it to Generate Video with a motion-only prompt. Download the two free sample images and follow the exact workflow first — then try it with a style reference of your own, which is where it becomes useful. For the desktop skills that turn a good generation into a finished image, our PRO Photoshop Learning Path is where to go next.

Quick Reference: Keys, Menus and Settings

• Open Boardsfirefly.adobe.com/boards, then create a new board
• Add your photosDrag the subject and the style reference onto the board
• Merge the stylesGenerate Image > pick a model (Gemini 2.5 Nano Banana in the video), eyedropper both images into the prompt, describe the edit, Generate
• More variationsClick Generate again without waiting; each click adds results
• Change the aspect ratioClick the result > the load icon (reloads the prompt), set Portrait, Generate, Place on Canvas
• Make a videoGenerate Video > pick a model (Veo 3.1 in the video), set the image as the first frame, describe the motion, Generate
• Save itThe download icon on the video; it comes to your computer with audio

Models and names are the ones on screen at recording; Boards adds and renames models often.