In the field of AI image generation, tools like DALL-E and Midjourney are dominating the throne - not just because of their high-resolution performance. The training of these models requires such massive investment and resources that inevitably lead to centralized services and pay-per-use access.

A new AI tool developed by the University of Surrey aims to reverse this trend and democratize high-resolution image generation by making it accessible to a wider audience.

This model, called DemoFusion, allows users to generate high-quality images without the need for subscription services or a very powerful computer. In fact, the system only requires a consumer-grade RTX 3090 GPU, which can be found in any mid-range gaming PC or Mac M1.

This AI is essentially a plug-and-play extension of the open-source model Stable Diffusion XL (SDXL), which generates images at a resolution of 1024×1024. DemoFusion is capable of achieving 4x, 16x, or even higher resolution increases - with just a few lines of code and without any additional training. The only trade-off, according to the team, is "a little more patience." We tried it at TNW, and it took about six minutes.

Left: Result generated by SDXL. Right: Result generated by DemoFusion. Image source: University of Surrey

To achieve these high-resolution results, scientists first generate low-resolution images and then enhance them using a process called progressive upsampling. This process works across image fragments to improve the details and resolution of SDXL.

"Our unique technology allows users to enhance their AI-generated images without the need for massive computing power or retraining the model," said Professor Yizhou Song.

"Digital art and images are a powerful medium that everyone should have access to - not just a few wealthy companies. That's why we made DemoFusion publicly available. We believe it can enrich our lives, and everyone should be able to use it."

Whether DemoFusion will gain enough traction to compete with giants like OpenAI's DALL-E remains to be seen, but its creation is an important step towards opening up the potential of AI image generation to the public and the broader tech community.