Base Model:T-noobnai3 V8 <a target="_blank" rel="ugc" href="https://civarchive.com/models/823566/t-noobnai3">https://civarchive.com/models/823566/t-noobnai3</a>Trigger: tomoshibiUSAGES:<a target="_blank" rel="ugc" href="http://1.Download">1.Download</a> the training dataset2.Copy all text prompts associated with the images containing the desired poses3.Remove any negative prompts from the copied text4.Add your preferred character's tags (e.g., the model's identifier) to customize the outputNegative Prompts: noise,watermark, too many watermarks, artist name, speech bubble, text focus, white borderTo ensure the accuracy of the generated images' poses, the prompts for each image in the training set were manually typed, rather than automatically generated by a tagging tool. Since the images required for training the LORA model must each have a resolution of 1024x1024, and to maintain the integrity of the image content, I added white borders to those images whose content would be incomplete after cropping. This ensured the completeness of the training images, which is why you may see white borders in some of the images in the training set.

Files with SHA256: c0a6f82244463b0f2bf8703453c2b844c3d592950c68d07bf8970104b5843c25