Images made with LORA weight 0.70 - 0.90
No trigger word needed
Sample images use highres fix. 640x360 x2, 0.5 denoise




















This model was trained on 150+ HD backgrounds from several movies/ova/shows.
The focus was without characters in frame at all. Roughly 10% or less of the training data had a character.
It produces beautiful backgrounds and environments with tons of detail.
I don't really recommend this model if you intend to make character shots/portraits.
Helpful Positives: no humans, scenery, landscape, grass, rock, stone, pebble, path, nature, forest, tree, ???? tree, leaf, plant, moss, fern, hill, mountain, horizon, mountainous horizon, river, day, sky, cloud, sunlight, sunbeam, outdoors, indoors, building, house, shiny, metal, crack, cracked wall, wooden floor, stone floor, still life, blurry, blurry background
This model was trained with only 16:9 widescreen images. generating 1:1 images works fine, but 16:9 may produce better results.
Images made with LORA weight 0.70 - 0.90
No trigger word needed
Sample images use highres fix. 640x360 x2, 0.5 denoise
1. 转载模型仅供学习与交流分享,其版权及最终解释权归原作者。
2. 模型原作者如需认领模型,请通过官方渠道联系海艺AI工作人员进行认证。我们致力于保护每一位创作者的权益。 点击去认领
