The LoRA was trained using 20 giantess images (initial version) I have generated using AOM3 last week.It is capable of creating female characters at/over building heights. The trigger word is "GiantessC" and best fits with "from above" or "from above" camera angles.I have tested the LoRA with the following models without merging: AnythingV3 - Good Any versions of Orange Mixes - Good Chilloutmix - Okay deliberate_v11 - Very Good Elysium_Anime_V3 - Okay Kenshi - Okay<hr />Update 01: It seems deliberate_v11 gives the BEST results in terms of raw model. (see images with crowded village)Make sure to have "height difference" or "size difference" to increase the chance of generating.<hr />Update 02: Replace images with newly generated ones using controlnet.<hr />Update 03 (12/09/2023): Last giantess related LoRA I have worked on.Camellia - OkayModels created by @GeneratedJudge - Okay (Works best with <a target="_blank" rel="ugc" href="https://civarchive.com/models/105941/tanoshimi-2d">https://civarchive.com/models/105941/tanoshimi-2d</a>)Perfect World - Okay<hr />Disclaimer: Depending on your model mix / prompt structures, outputs may vary and the chances of giantess occurrences may not be 100%. (Roughly 70% work on my test cases).<a target="_blank" rel="ugc" href="https://twitter.com/Claire_Miyamoto">Twitter</a>

AOM3 Based Giantess LoRA (Sunset)

Trained on 50 images (dataset), using 768x768 instead of 512x512.

Files with SHA256: 57d2eb8540de7007cea9d357663174ef8c48bec08486593b1a83d4df1b3cf389