我是社区的新手，欢迎大家随意使用。这个Lora可以画出像素人物和场景，偏向于32x32和RPG像素游戏的风格。Lora基于100张不同的32x32像素画在SD 1.5上训练，使用SD 1.5和Anything 4.5都有不错的结果。I'm a newcomer to the community, and everyone is welcome to use my services freely.Lora can draw pixel characters and scenes, with a style that leans towards 32x32 and RPG pixel games.Lora is trained on 100 different 32x32 pixel drawings using SD 1.5, and has good results with both SD 1.5 and Anything 4.5.我在Huggingface上传了不同epoch的结果，<a target="_blank" rel="ugc" href="https://huggingface.co/skyatmoon/MyPixelArtLora">https://huggingface.co/skyatmoon/MyPixelArtLora</a> , 欢迎参考下面的epoch-lora_weight图选择想要的风格和权重，一般情况下权重为1就可以了I have uploaded results from different epochs on Huggingface, <a target="_blank" rel="ugc" href="https://huggingface.co/skyatmoon/MyPixelArtLora">https://huggingface.co/skyatmoon/MyPixelArtLora</a>. Please refer to the epoch-lora_weight chart below to choose the desired style and weight. In general, a weight of 1 is sufficient.<img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/9046c6c6-521c-4a3b-f982-edc505b54b00/width=525/9046c6c6-521c-4a3b-f982-edc505b54b00.jpeg" />使用不同的采样器在不同权重下结果差异较大，可以根据下面的Sampler-Weight图选择，一般使用DMP++ 2M Karras就可以了Using different samplers at different weights can result in significant differences. Please refer to the Sampler-Weight chart below to choose the appropriate sampler. In general, using DMP++ 2M Karras should be sufficient.<img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/4eb96574-b62e-4959-9299-917211570c00/width=525/4eb96574-b62e-4959-9299-917211570c00.jpeg" />Active Tag: basepixel

Pixel Art style Lora

First version for 20 epoch training

Files with SHA256: 8b7ea7690b0cc355e561475483452e61f537ed875604e0bb38d45ec07039f06f