<aside> <img src="https://choosemuse.com/cdn/shop/t/49/assets/muse_favicon-1000x1000.svg?v=31343724533644752771712001645" alt="https://choosemuse.com/cdn/shop/t/49/assets/muse_favicon-1000x1000.svg?v=31343724533644752771712001645" width="40px" /> Disclaimer → Muse is a product by Interaxon, the company I work for as a Data Analyst. The project is about the Stable Diffusion model, it just happens to be of pictures wearing our product.
Everything in here is my own opinion and I have permission to share this project. The resulting model was never used in production and was created for fun.
</aside>
Muse S Elephant
Muse S Elephant
Muse S Dog
Like many others, I’ve become obsessed with AI. There are so many cool projects to pursue, and I have so many unfinished projects. This project is special becuase how great it turned out, and how surprisingly little effort it was. For context on who I am and why I did this - data analyst/data scientist who spends a lot of time in hobby projects. Also I have an RTX 3090 that lets me run some large (as far as consumer grade goes anyway) open source & pretrained AI models, such as those found on HuggingFace.
All the tools used here are free and can be found here:
So, why did I do this?
In the early days of AI I recall people being excited about image generation capabilities. In company or team meetings where AI generated images were used to represent the company or its products, it was always done for fun & nowhere near represented the actual product. I knew it was possible to fine-tune models, so the moment my eyes gazed upon the Muse Marketing Press Kit I was compelled to try it out myself, as I was already deep into Stable Diffusion for hobby projects.
The Muse Press Kit contains ~50 high quality images of people using our product the Muse 2 S (sleep version). This is somewhat just an experiment with a goal of creating new, images in the style of the press kit. New people, new backgrounds, etc. All wearing the same product on their heads.
