This is the Flux version of the Goonstyle Lora. It features a trashpunk aesthetic that is character-focused and has a wide range of environments, backgrounds, color, lighting, and poses. It emphasizes strong character design in rugged scenes with strong lighting contrasts involving vibrant light and shadows. This Lora features a dataset that is more than twice the size of my Pony Lora and is trained at 1024. 230 images were each manually captioned to ensure accuracy of pose descriptions, clothing, and setting.

The model is likely to respond well to descriptions of clothing that especially focuses leather (vest, jacket, trenchcoat, pants, belt, boots, wrist bands, bracers, etc) as well as descriptions of tears, rips, studs, spikes, and patches. It will do well with descriptions of tattoos (neck, forearm, sleeve, arm, hand, etc). Jewelry was included in the prompts (ring, watch, necklace, etc).

The aesthetic is broken into multiple types that feature industrial-style settings, Victorian-style settings, and modern domestic-style settings. Each has a focus on clutter and trash in the scene such as paper, bottles, and clothing. The Lora will respond well to prompts about damaged or distressed walls, furniture, and scene objects.