Krea-GiantessLora v1
Trained with 81 images. Mixture of photorealistic, anime, drawing, AI gen. Contains some futanari training data and data with muscular women that might bleed into generations on high weights.
Recommended weights:
photorealistic 0.7-1.4,
anime-flat: 0.8-1.2 (smaller range for better results)
3drender: 0.7-1.3
drawing: 0.85-1.2
No official trigger word, but use giantess to trigger it (was used as first word in dataset). Also responds to "mega giantess (scale)" and "giga giantess" to some degree. Use with supportive detailed descriptions on prompt. Lora should help with interaction and prevent perspective issues.
If you want futanari content use another lora to bring it out and use lora on higher weights to support it better. (haven't tested this too much yet)
DO NOTE that this is my first lora that I have published, so I can't answer to too hard questions you might have.
Description
Giantess V1
base version for lora
FAQ
Comments (2)
What trainer you did use?
Might as well train LoKr to avoid bleeding because of multiple styles. Ai-toolkit is preffered way
Thanks!
soz for little late answer, musubi-tuner, followed this guide to most part with some adjustments: https://note.com/sepiablue/n/nbc355cc7e114?hl=en worked out prompts mostly with gemini with personal edits, some more "harder" content with grok for dataset