MiladTower

It is designed to reproduce Milad Tower accurately using the trigger MiladTower. The training set includes distant views, close-ups of the upper structure, low-angle shots, and several 3D renders. The goal is to preserve proportions, structural form, surface details, the pod, and the top geometry.

Background changes, lighting variations, time-of-day shifts, and different artistic styles are supported without losing the tower’s identity. For tight close-ups, place the trigger at the start of the prompt. For wide shots, combine the trigger with scene description.

Trained at 768 resolution on Flux-dev. Works reliably for realistic, cinematic, and architectural outputs. Recommended strength: 0.8–1.2. Trigger must be used exactly as MiladTower.

This LoRA is an early version and open to feedback.