<a target="_blank" rel="ugc" href="https://patreon.com/psoft"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/8dcbc56a-15e9-429e-9b36-c2bc22594678/width=525/8dcbc56a-15e9-429e-9b36-c2bc22594678.jpeg" /></a> Support my work on <a target="_blank" rel="ugc" href="https://www.patreon.com/psoft">Patreon</a> and <a target="_blank" rel="ugc" href="https://ko-fi.com/psoft">Ko-Fi</a> and get access to tutorials and exclusive models. Rate and leave a like if you enjoyed it, and follow for new models!<hr />This is still an experimental NTR LoRA. It works but it's hard to prompt. Don't expect easy results and don't take it too seriously - yet.It's a complex concept / scene with different compositional elements and multiple characters and poses. So it still needs some work. The dataset is insufficient so it was difficult to strike a balance between good composition for generalization without affecting style or overfitting. This one might be slightly overfitted, so weight 0.9 is probably better. And you need to use hires fix for your final image, otherwise it won't look as good. Some aspect ratios work better than others.The trigger word is caught, the rest is about prompting the position and the characters. Please refer to my example images, as they are the best place to start.For some poses/angles/aspect ratios you will have to generate a bunch of images before you get a good one. I usually do batch 8 to get 2-4 good results. Yes, my example images are cherry-picked. But this is not a single pose or a single outfit. It is a concept that mangles many components together and it's very tough without a proper dataset.Anyway, I still hope you can produce some nice generations out of it.

Files with SHA256: 5d33eb6a6bc096f1bae5b8ac3f06fc007f834106b3cb556d5ee8eaeb0c571e13