This Lora was a tricky one to make since she has a large dataset and that why choosing the right images took a lot of time. After searching through every inch of the internet, i dense it down to 183 images. Trained on 768x768 reso. Use weight between 0.7 ~ 0.8 . I don't think this model is perfect by any chance but it a decent one. If you like the model please consider leaving a review. Trained for 4 epoches but the forth one is overfit so decided to publish the 3rd Epoch.I was going to fix the issues but since i'm too lazy to do it here are the known issues :<ul><li>Her hoodie can be whacky sometime especially clothes manipulation such as windy effect or open clothes etc.</li><li>Her blue pants is not recognized very well resulting in blue jeans pants. (probably need more training for the lower half)</li><li>Can't produce fishnets really well. At least it's acceptable. So, Good Luck trying.</li><li>On her alternate hairstyle(side ponytail),can appear on both side. If that happen Keep generating. you'll eventually get the right one or the left XD. </li></ul>This Lora Include 2 versions the Original Shippuden Outfit and the Alternate Hairstyle from the onsen Scene. Both versions can be used interchangeably.Basic Prompt for Shippuden Outfit: <code>masterpiece, absurdres, 1girl, hinata\(shippuden\),solo,blush, hooded jacket, headband around neck, konohagakure symbol, fishnets, blue pants, looking at viewer, outdoors, sunlight,smile,cowboy shot</code> Basic Prompt for alternate Hairstyle: <code>masterpiece, absurdres, 1girl, hinata\(shippuden\), solo, naked towel, large breasts, grabbing own breast, alternate hairstyle, side ponytail, looking at viewer, blush, embarrassed,</code>If you want to replicate the sample images here is the model that i used : <a target="_blank" rel="ugc" href="https://huggingface.co/andite/anything-v4.0/blob/main/anything-v4.5.safetensors">anything-v4.5.safetensors</a>

Files with SHA256: a23595707b999cc0f73a16dffe7e0d5efb8a8af6d8c0bfc16e27d02671658892