<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/cadb0da0-3f97-47d4-a852-ee2df3a2bfe8/width=525/cadb0da0-3f97-47d4-a852-ee2df3a2bfe8.jpeg" />Potassium rich LoRA infused with the likeness of <a target="_blank" rel="ugc" href="https://twitter.com/Banano_Chan/">Banano Chan</a>. The digital waifu embodiment of <a target="_blank" rel="ugc" href="https://www.banano.cc/">Banano</a>, a feeless and super fast meme cryptocurrency.<ul><li><a target="_blank" rel="ugc" href="https://tensor.art/models/619769699548517287?source_id=nz-_pVnmkkW1ofcjb3fx-hYl">Use this LoRA on TensorArt (SD 1.5, Pony, Flux)</a></li><li><a target="_blank" rel="ugc" href="https://glif.app/@pbuyle/glifs/cm295zz7j0009zgyf5g2grwlr">Use this LoRA on Glif (Flux)</a></li><li><a target="_blank" rel="ugc" href="https://pixai.art/model/1622843490460652098">Use the LoRA on PixAI (SD 1.5)</a></li><li><a target="_blank" rel="ugc" href="https://www.seaart.ai/models/detail/187c1f4868c92659d27bc190d2bc95ed">Use this LoRA on SeaArt (SD 1.5)</a></li><li><a target="_blank" rel="ugc" href="https://replicate.com/pbuyle/banchan-flux">Use this LoRA on Replicate (Flux)</a></li><li><a target="_blank" rel="ugc" href="https://novita.ai/?ref=mjnkzji&amp;utm_source=affiliate">Use this LoRA on Novita.ai (SD 1.5, Pony) - Accessible in the Playground or API, no direct link.</a></li></ul>Note that while the SD1.5, SDXL and Pony versions all use <code>banchan</code> as trigger word, the Flux version uses <code>b4nch4n</code>. The Flux version is also less aware that Banano Chan is a woman, so you need to provide that information in your prompt. Either directly with keywords such as <code>woman</code> or <code>girl</code>, or by using pronouns when describing her outfit, pose, etc. Training was done using only anime/cartoon illustration and this leaked into the training. Without additional LoRA or an fine-tuned opinionated base model it is hard, but not impossible, to get other styles, incl. photography. When doing realistic style, the hair colors tend to go away if not included in the prompt.

Banano Chan (Flux, Pony, SD1.5, SDXL)

Trained on Flux.1 [dev] with 20 images (a mix of originals and generated pictures using the SD1.5 and Pony versions) captioned with <a rel="ugc" href="https://huggingface.co/spaces/fancyfeast/joy-caption-pre-alpha">JoyCaption</a>.

Files with SHA256: 2bab5c8dba0846c10f78a38bd91aae59b1f57ae041817948ace77eab99983262