Another experiment to explore alternate use cases of Generative AI - in this case, the ability to simulate potential environmental impacts in existing landscapes/images; as well as ability of style LoRA to extrapolate from the diverse imagery from its training set of photographs of degraded landscapes that exhibit scars of human/industrial activity.<h3 id="lora-weights-mike54eqt">LoRA Weights</h3>Due to diversity of the imagery in the training set, the optimal weight for your image will likely vary. In the sample below of "forest with city in the background", you can see the LoRA influence on this particular image is between 0.4-0.6.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/b146d803-7ec5-43de-833f-a61e98531f9c/width=525/b146d803-7ec5-43de-833f-a61e98531f9c.jpeg" />

EcoDamage

V1 uses autotagged photographs, which will be improved upon in later versions.

Files with SHA256: bafcc0cc670cd76e4303eb7cac770aa7c563d2c32551802b4d3d235449fe2b1a