A LoRA for NoobAI, trained on semirealistic SFW images generated with GPT-Image (which is a proprietary model). This style has distinctive traits: 1. Smooth, CG-like shading and lighting for 3D-like shapes, without linework but with manga-like anatomy; 2. Highly-ornated compositions, with trims added randomly on surfaces and objects; 3. Vibrant colors, with use of complementary colors for shadows and reflected lighting; 4. Curved shapes, esp for such details as hair strands and clothes folds. Use gp1st as a trigger word.Update: added version 1.5 which is more stable (due to more diverse samples in training data) and has learned more of specific features for the target artstyle.

Files with SHA256: a0a805ead4c88da99af2ce720d9f0b6115876e1121b6ef56d264cae65092d760