这是做烟花场景的lora，训练图用了51张训练图（训练图大多是midjourney生成的图）,底模用的是pasteldiffusedmix_v20，尺寸是512x768,描述词简单描述就好了，例如天空, 烟花, 无人, 风景, 云, 户外, 夜, 云天, 倒影, 夜空, 星（天空）, 水, 日落, 地平线, 星空, 海洋...我试了其他偏二次元风格模型大多也合适，权重调0.5-0.8左右，不宜太高；我个人觉得有丢丢鸡肋（。・＿・。）ﾉ，大概就只能生成烟花图，做做壁纸就还是可以的；The following is machine translatedThis is the lora for the fireworks scene, 51 training maps were used (most of the training maps were generated by midjourney), the base model is pasteldiffusedmix_v20, the size is 512x768, and the base model is 512x768.The base model is pasteldiffusedmix_v20, size is 512x768.The descriptors are simple.For example sky, fireworks, no one, landscape, clouds, outdoors, night, clouds, reflection, night sky, stars, water, sunset, horizon, starry sky, ocean...I've tried other secondary style models and most of them are suitable, the weights are adjusted around 0.5-0.8, not too high;Personally, I think it's a bit weak. ___・・.) Use, probably can only generate fireworks map, do wallpaper will still be able to do;

fireworks

这是做烟花场景的lora，训练图用了51张训练图（训练图大多是midjourney生成的图）,底模用的是pasteldiffusedmix_v20，尺寸是512x768,描述词简单描述就好了，例如天空, 烟花, 无人, 风景, 云, 户外, 夜, 云天, 倒影, 夜空, 星（天空）, 水, 日落, 地平线, 星空, 海洋...我试了其他偏二次元风格模型大多也合适，权重调0.5-0.8左右，不宜太高；我个人觉得有丢丢鸡肋（。・＿・。）ﾉ，大概就只能生成烟花图，做做壁纸就还是可以的；The following is machine translatedThis is the lora for the fireworks scene, 51 training maps were used (most of the training maps were generated by midjourney), the base model is pasteldiffusedmix_v20, the size is 512x768, and the base model is 512x768.The base model is pasteldiffusedmix_v20, size is 512x768.The descriptors are simple.For example sky, fireworks, no one, landscape, clouds, outdoors, night, clouds, reflection, night sky, stars, water, sunset, horizon, starry sky, ocean...I've tried other secondary style models and most of them are suitable, the weights are adjusted around 0.5-0.8, not too high;Personally, I think it's a bit weak. ___・・.) Use, probably can only generate fireworks map, do wallpaper will still be able to do;

Files with SHA256: 02581e21d0080a1f1c012fb90fa6e919d66059fd2e0d136e5e9d8e113ce0c4f6