Type | |
Stats | 1,489 2,282 |
Reviews | (334) |
Published | Oct 28, 2023 |
Base Model | |
Hash | AutoV2 26056A35FA |
这是做油画风格lora,适合做油画风格场景,训练图用75张训练图,
底模用的是pasteldiffusedmix_v20.safetensors,尺寸是768x512,
我尝试了一些二次元风格的大模型,有些合适,有些不合适,权重看着调;也尝试类似revAnimate,etheerRealMix这些2.5D风格的大模型,权重要提高一些;
训练图大多是有光感的,所以生成的图也一般带有光感氛围,总体生成的色感也还可以;
由于训练图是以户外风景居多,生成户外风景效果相对好些,生成人物效果不是很理想;
The following is machine translated
This is to do oil painting style lora, suitable for oil painting style scenes, training charts with 75 training charts, the
The bottom model is pasteldiffusedmix_v20.safetensors, size is 768x512.
I tried some secondary style big model, some suitable, some not suitable, the weights look at the adjustment; also try similar revAnimate, etheerRealMix these 2.5D style big model, the weights will be increased a little;
Most of the training charts are light, so the generated charts are generally with light atmosphere, the overall generation of the color sense is also okay;
As the training map is mostly outdoor scenery, the effect of generating outdoor scenery is relatively good, and the effect of generating characters is not very ideal;