Exploring Stable Diffusion AI for 3D Visualization

As part of our ongoing exploration of emerging technologies, I recently tested Stable Diffusion AI directly within our 3D application. My goal was to assess its potential impact for our 3D visualization team and explore its creative value for designers.

To start, I created a straightforward scene with a few complex elements for initial testing. Below, you’ll find the render produced using Fstorm—without any AI intervention, just a pure render.

With my prior experience in Stable Diffusion, including training our own models, I was already familiar with how SD works. After some fine-tuning, I’ve managed to achieve what I believe are some impressive results.

Currently I think the best use is in adding intricate details to existing renders, enhancing textures, and speeding up the look development process, particularly when it comes to time-consuming finishes.

The key moving forward is how to use this tool responsibly and skilfully, ensuring it complements and elevates our team’s work.

Below, I’ve shared some of the results, showcasing a range of outcomes—from mild to wild.