Adobe MAX 2025 Tarot Card Project
This past year's Adobe MAX preconference session was called the Magical Mythological Menagerie and featured Adobe Firefly AI solutions to create a custom tarot card and then animate a related video.
The source image for my demonstration was captured in the previous year's preconference where I was the Magician for a day...
From this single source image I then used Adobe Firefly Text-to-Image with various styles and prompts to generate my starting outputs. My goal was not to have AI 100% generate my card. I still wanted to have full creative control and composite using existing tools.
Each output was a varying generation where I edited my prompts and style to give me more options than I could possible need. From here I brought each element into Photoshop to generate my composite.
Almost every generated result was used in the composite from adding atmospheric elements like the colored dots in the background and coat, to the Neo Tokyo backdrop.
Darker elements like the ink washes and shadowed trees brought a moodier feel to the overall composition. The face still needed a lot of work as it did not look like me.
Using the latest Adobe Firefly third-party model (at the time) Google Gemini 2.5 I was able to focus in on my face and generate a similar look and feel using the composite image as a style source.
The original output (left) and the final output (right) with my glasses added.
Below is the final composite for the print. Using this image, I then starting looking at how to incorporate this artwork into my Tarot card design.
Starting with a supplied graphic used by the students in the class, I did the same exercise as with the portrait image - generate various outputs with different styles and prompt edits. Below is an example of the "steampunk" option I focused on.
And so this was the final Tarot card for print compositing the portrait image with the custom frame elements.
The second part of the project involved animating your Tarot card. Students used Firefly Text-to-Video aspects including various third-party models to produce video clips and audio. They then combined them to output a final video for display.