Updated: Feb 2, 2026
concept
This LoRA is designed to adjust the environmental density of your image.
Trained on 600+ images of highly dense environments with minimal captioning, it works like a slider.
Strength might vary for your base models, works well for Illustrious, and Flux. QWEN seems a little less obliged, but it still works - you may need a little experimentation on your prompting and strength.
Tested from -2 to +2 strength.
----
Added Z Image Base. It may be better between -1.5 to +1.5 strength. Let me know how it goes :)
----
Added F2K. Again -1.5 to +1.5 strength seems to be an acceptable range!