So, uh, I had this dataset made up of captioned images from the game Third Crisis (NSFW), and I rented a GPU on RunPod, and I chose Animagine XL V3 Base (license) as checkpoint to train on, and I selected the AdamW optimizer, and I pressed train, and my budget went down, and this is the result. If you like the vision behind it, consider supporting me via Ko-Fi, which might enable me to make some improvements. Thank you for your attention.