About

I'm Dan, and this site combines my passion for AI-generated art and my cat Psy.  

Origin

I first became aware of AI capable of generating images from text in early 2021 when OpenAI revelead their breakthrough technology called DALL-E. I was blown away by the demos, and despite studying an extensive amount of AI in college, could hardly believe what I was seeing was true. Shortly after, open-source projects began to pop up emulating the technology behind DALL-E. As you can see below, the earliest iterations produced shapes and colors vaguely resembling the prompt. Nonetheless, I was captivated and closely followed advancements ever since.

Emergent Coherence

After several months, the open-source text-to-image projects evolved to the point of generating coherent images. "Black Cat" became my baseline prompt to compare technologies and has stuck with me ever since. These tools were great for generating artistic or abstract style works, but fell short in the photorealism department.

Major Breakthrough

In mid 2022, OpenAI once again collectively blew our minds with the announcement and demos of DALL-E 2, capable of producing photorealistic images. Not long after, a comparable open-source technology Stable Diffusion emerged and remains a core component of the latest developments.

Moving Forward

While I don't believe future advancements will be as visually striking as the jump from DALL-E to DALL-E 2, there are still exciting discoveries happening on a regular basis. Recently, the ability to train and merge models has become widely accessible and opens the door to a much deeper level of customization and personalization. I plan to continue following the topic closely and posting images as the technology and my understanding of it improves.