天若有情天亦老，老子也是冬马党。大家好，我是chosen鲷，这次给大家带来的是东马和纱人物lora。值得一提的是，这次的人物lora并不是完全还原原作，因为我的训练集不都是白色相簿2的cg图，不得不说十年前的gal图cg那个画风在今天看实在是有些过时了，训练集里有很多是p站上的同人图，然而东马的同人图格外的少，所以我还放了一些用之前深崎暮人lora重绘的动画截图（没错，我训练深崎暮人lora就为了这个，属于是为了这点醋包的饺子），然后我还加入了一些鬼月的御姐图，虽然听起来这俩完全不沾边，但是我隐约觉得鬼月的御姐脸型放在东马身上肯定不违和，于是便训练出了这个lora。画风我还是很满意的，是我很喜欢的脸型，并且还保留了原有人物的那种神韵（不过这个就见仁见智了）。然后我训练的时候脑子抽了，忘记删掉black hair和blue eyes这俩tag了，导致每次写的时候都得加上这俩才能保证出东马，否则头发颜色会不对。然后可以画一些和雪菜的贴贴图，例图中那几个我是用ctrlnet的softedge模式引导做出的。推荐的tag：(masterpiece:1.2), best quality,pixiv, 这里写上你想要的内容 &lt;lora:Touma Kazusa_20230716214229:1&gt; Negative prompt: EasyNegative, badhandsv5-neg,Subtitles,word, Steps: 32, Sampler: DPM++ 2M Karras, CFG scale: 9, Seed: 683254128, Size: 400x800, Model hash: 684006e5fe, Model: chosen Irises-mix, Denoising strength: 0.5, Clip skip: 2, ENSD: 31337, Hires upscale: 2, Hires upscaler: R-ESRGAN 4x+大模型方面，首推irises，我的其余几个chosen mix，impasto也行，常见的一些大模型基本都行，这lora兼容性蛮好的。感谢大家的下载使用，如果喜欢的话欢迎关注我，我的bilibili pixiv ID均为chosen鲷，欢迎大家关注我的频道<a target="_blank" rel="ugc" href="https://space.bilibili.com/3867754?spm_id_from=333.1007.0.0，https://www.pixiv.net/users/40911695">https://space.bilibili.com/3867754?spm_id_from=333.1007.0.0，https://www.pixiv.net/users/40911695</a> 谢谢。Hi, I'm chosen 鲷, and this time I'm bringing you Touma Kazusa character lora.It is worth mentioning that the character lora is not completely restored to the original, because my training set is not all cg pictures of white album 2, I have to say that the gal picture cg style of ten years ago is really outdated in today's view, there are many fan pictures on p website in the training set, but Dongma fan pictures are particularly few. So I also put some animated screenshots that I redrew with the previous Fuzaki Twilight lora (yes, I trained the Fuzaki Twilight lora for this, belonging to the vinegar dumplings), and then I also added some pictures of the ghost moon, although it sounds completely different from the two, but I vaguely feel that the ghost moon's royal sister face is definitely not compatible with the East horse, And so this lora was trained. I am still very satisfied with the painting style, it is the face shape I like very much, and it still retains the charm of the original character (but this is a matter of opinion).Then I got crazy when I was training, and I forgot to delete the tags of black hair and blue eyes, so I had to add them every time I wrote to ensure that I would be in Dongma, otherwise my hair would be the wrong color. Then you can draw some stickers with snow cabbage. The ones in the example picture I made with ctrlnet softedge mode.Recommended tag:(masterpiece:1.2), best quality,pixiv,Write what you want here&lt;lora:Touma Kazusa_20230716214229:1&gt;Negative prompt: EasyNegative, badhandsv5-neg,Subtitles,word,Steps: 32, Sampler: DPM++ 2M Karras, CFG scale: 9, Seed: 683254128, Size: 400x800, Model hash: 684006e5fe, Model: chosen iriss-mix, Denoising strength: 0.5, Clip skip: 2, ENSD: 31337, Hires upscale: 2, Hires upscaler: R-ESRGAN 4x+In terms of large models, the first push irises, my remaining chosen mix, impasto is also OK, some common large models are basically OK, this lora compatibility is quite good.Thank you for downloading and using, please follow me if you like, my bilibili pixiv ID is chosen Snapper, Welcome to focus on my channel, <a target="_blank" rel="ugc" href="https://space.bilibili.com/3867754?spm_id_from=333.1007.0.0">https://space.bilibili.com/3867754?spm_id_from=333.1007.0.0</a>, <a target="_blank" rel="ugc" href="https://www.pixiv.net/users/40911695">https://www.pixiv.net/users/40911695</a>, thank you.

Files with SHA256: 88b0e13f22dac55bee9f548e976f019ff042823c31ce3b38bc67ba38c98d8899