Font Textures with Generative AI

This project explores the use of generative AI to create font textures for 3D printing. The workflow uses the new Adobe Firefly (Beta) text effects feature to generate 2D images that are converted to 3D models for 3D printing.

• Create Image: Use Adobe Firefly (Beta) - Text Effects to generate image. https://firefly.adobe.com/generate/font-styles (note: My academic institution provides access to Adobe products for students and staff.) Enter the text to be created and a written description of the text effect desired. In the examples provided, the written descriptions were 3D printing in black and white and Japanese woodblock waves in black and white. In my experimentation, I found a black and white image created better results for me when converting from 2D to 3D. Other options include different fonts and shape matching as well as 4 design iterations.
• Download Image: Download the image. The image will download as a PNG with transparent background and an Adobe Firefly watermark. Crop image to remove watermark.
• Convert 2D Image to 3D: Open PNG image in Ultimaker Cura. I used version 5.2.1. The Convert Image dialog will open. I used the following settings to generate example 2. Height 2.5mm, Base 0.4mm, Width 120mm, Depth 111mm,  Lighter is higher, Color Model Linear, 1mm transmittance 50% and Smoothing 2.
• Export STL: Export STL from Ultimaker Cura. File → Export → STL.
• Slice STL for 3D Printing: I printed examples on Prusa Original Mini so imported the model into PrusaSlicer version 2.6.0 using the 0.15mm Quality presets. I ignored errors about open edges and sliced the model. (Repairing via NetFabb in PrusaSlicer did not work for me.) I explored the layer preview and added a manual color change at a point that provided good separation between base and texture.

I think this is an interesting experiment in the use of generative AI to create font textures. I would be interested in seeing examples of this technique from other users as well as other workflow ideas.