LoRA to create giantess content.

Note that there are 4 example images for each version. If you can not see all examples you need to log in. If you still can not see all images, go to Settings and then Content Moderation and allow Mature Content. None of the images are mature by any stretch of the definition, but Civitai will sometimes tag an image as such if as much as an ankle is shown.

gb4:

Trained using over 11000 colleges with corresponding tags taken from Giantess Booru.

bg13:

For this new version I went through all the images used in gb4 and removed the ones that were low or mid quality, leaving only 2252 collages. I also added 259 high quality collages and manually tagged them.

I think that at this version works as well or better than the previous version. However, in some cases a prompt that worked well with gb4 might work less well with gb13. In this case try tweaking the prompt slightly until it works better.

Normally strength 0.8 seems to work best, but you can try to lower it. I noticed that a lower value tend to produce a smaller giantess.

This version of the LoRA was trained using the CyberRealistic 4.1 as the base model. I highly recommend using that model for this version of the LoRA.