Refined Taxonomy

'Refined Taxonomy' began as an artistic experiment in model refinement. Using StableDiffusion 1.5 (trained at 512x512,) for high-resolution generations tends to create a unique effect: larger resolutions result in unexpected patterns, such as multiple heads and limbs. However, by starting with SD 1.5 and then transitioning to the new SDXL refiner model after just a couple steps, we can achieve clearer, higher-quality images. This process offers a significant aesthetic improvement while preserving the original abstract compositions.

This LoRA intertwines traditional elegance with a contemporary edge. The technology employed mirrors this duality, transforming rough initial drafts into polished final pieces.

'Refined Taxonomy' is a blend of past and present, showcasing animals in aristocratic garb through an AI that has learned from a biased dataset. The project seeks to embody 'refinement,' yet one must acknowledge that this prompt focuses heavily on terms seeped in classism and Eurocentrism, overlooking the rich variety of sophistication in other cultures.

When engaging with these portraits with a critical eye, considering the AI's interpretive limits and the impact of the biases within the tools and creators. 'Refined Taxonomy' offers an opportunity for conversation about inclusivity and honoring all cultural narratives in the realm of creativity.

They're also just fun and pretty to look at.