Recommended weight： 1
Trigger word：
belly punchingNote: Reopening onsite generation, but please be mindful of the content and its boundaries.
It's unfortunate, it seems like it has been unpublished again.
After considering the possible reasons, I've decided to only keep PG-rated example images. I believe that this model isn't solely for producing violent or abusive images;
it can be used to recreate some iconic anime fight scenes, such as DIO vs. Jotaro. However, I also understand Civitai's concerns, so I'll try only PG POSTs and disabling live images on Civitai. Thank you.
Feel free to leave feedback in the comments below!
______________________________________________________________________________________________
建議權重：1
觸發提詞：
belly punchingNote: 重新開放生圖，還是要請各位注意尺度喔。
很遺憾，似乎又被Unpublished了，考慮可能的原因過後，決定只保留PG級的返圖。
我認為這個模型本身不是只有生產暴力或是虐待圖片用途，而是可以拿來重現一些經典的動漫打鬥畫面，例如DIO對上承太郎的場面。 但是我也理解C站擔心的隱憂，因此嘗試看看只能貼PG POST和關閉C站生圖，謝謝。
歡迎底下留言反饋!
Description
FAQ
Comments (17)
this looks better then previous belly punching LoRAs i saw
铁块！
我可太爱这个lora了
Any chance for a 1.5 release of this LoRA?
This has been helpful in boxing gens.
awesome work
Any chance of this one coming back?
Unfortunately, this is the current situation for now. However, I will consider uploading the ILXL version and enabling online generation. If that version works without issues, I will reconsider it later.
@zoca yess. Please make an Illustrious version. Or if this is requires much time please share the dataset and I will train it
@zoca NVM, saw that you alerady published an IL XL version. Thanks mate!
Any plans of a version for SD1.5?
There is already a trained model, but due to some issues that occurred before, the upload plan was put on hold. I'm still considering it at the moment.
可以请教一下训练集是如何收集的吗?
权重拆分的很好对低模的性能影响极小并且在很多低模上表现都很好, 不知训练有何技巧?
如果你是剛接觸訓練的話，可以先去b站上看一些大佬的教程，或是去看相關論文，會比較有幫助。至於訓練集準備的部分我很難給你一個標準答案，只能講個模糊的通說，就是品質越高越好，數量就看設備條件允許的程度。
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.