The trigger word is velkhana_alpha

Good at 0.7-1 weight

Tips:还得再说明一点,之后我可能不会再更新这个lora了,我希望压缩包中的训练图片可以帮助有心之士进行更新,我得训练下一个装备lora

I should also mention that I probably won't update this LoRA in the future. I hope the training images included in the archive can help those who are interested in improving or updating it. I need to move on to training the next armor LoRA.

V1:训练素材来自游戏截图,通过GPT和Gemini进行编辑以符合训练要求,最终训练用图片数量在39张,最终输出图片可能会受训练素材的影响,建议使用画风lora进行控制,

武器的触发词:greatsword,huge sword.但是这生成的效果并不好,因为我并没有对武器进行专门训练,它有可能生成还不错的武器形状

这是我第一次训练lora,对于我来说最难的就是素材的收集,网络上并没有velkhana alpha的高清展示图,只能靠我自己在MHW和MHR中进行截图并交给大语言模型进行编辑,不过还好最终结果还算差强人意,这对我来说是好的开始

The training materials were sourced from game screenshots and edited with GPT and Gemini to meet the requirements for training. The final training dataset consisted of 39 images. The generated images may be influenced by the training materials, so using a style LoRA is recommended for better control of the overall art style.

The trigger words for the weapon are: greatsword, huge sword. However, the generated results are not very good because I did not specifically train the weapon. It may still generate a decent weapon shape in some cases.

This was my first time training a LoRA. For me, the most difficult part was collecting the training materials. There were no high-quality showcase images of the Velkhana Alpha armor set available online, so I had to capture screenshots myself in MHW and MHR, then use large language models to edit them to meet the training requirements. Fortunately, the final result was still fairly satisfactory. This is a good start for me.