This LORA for background.
Postwar Ruins, a completely devastated postwar urban street extends deep toward the center background, bordered on both sides by severely damaged multistory reinforced-concrete buildings. In the left foreground, a dirty, torn brown upholstered sofa covered with dust and debris rests among fractured concrete slabs. Along the left side, cracked reinforced-concrete columns support partially collapsed floor structures with rusty, bent reinforcing bars hanging from broken edges. In the right foreground, large rough fractured concrete blocks lie among rusted steel rods and crumpled corrugated metal sheets. In the right middle ground, crushed rusty metal panels and twisted structural steel form a dense debris pile. The central dirt roadway is dusty, uneven, and scattered with small concrete fragments, leading directly toward distant ruined buildings. On both sides, shattered walls, exposed floors, broken balconies, and open reinforced-concrete structural frames line the street. The concrete surfaces are deeply cracked, chipped, dusty, and heavily weathered, with exposed aggregate and rust-stained reinforcement visible throughout. Warm muted daylight illuminates the dusty brown-gray field of ruins.
I created datasets images with AI (ChatGTP) based on my own prompt. LORA was created with fluxgym (RTX-4090).
Description
I created the 5 datasets images with AI (ChatGTP) based on my own prompt. LORA was created with fluxgym.
I trained this LoRA using a prompt-addressable implicit spatial representation. It functions as a text-based scene-control system: location + sub-theme + object + spatial relation + viewpoint + interaction.














