I did Pombon already so I thought I might as well do Gecqua (DW Browt will be coming too). From my testing, due to the sheer number of NSFW images in the dataset, you may need to add penis, pussy, balls, anus to your negative prompt to get this to gen SFW. This was trained on 80 images with 6000 steps, but even then it can still sometimes mess up. Gecqua must be a hard character for AI to understand. I will probably make a v2 someday.