This model generates big anime girls sitting on top of or hanging over cities and landscapes. It's fine-tuned for NSFW images and has a large NSFW preference.

This model was trained on 150 mostly NSFW images using Dreambooth and then merged with the "Giantess | Concept" LoRA from @LeonTalon which further refined the results.

Recommended settings for good results:

Sampling method: DPM++ SDE Karras

Sampling steps: 30 steps

height and width: 512x512

CFG Scale: 7

Clip skip: 1

Further fine-tuning might give better results depending on your preferences

Higher resolution images (highly recommended if your PC can handle it!):

Hires. fix: enabled (scaling from 512x512 to 1024x1024)

Upscaler: Latent (bicubic)

Upscale by: 2

Denoising strength: 0.65

Hires steps: 0

I didn't do allot of experimenting with this other settings will probably give better results

Recommended networks for better results:

Easy Negative

negative prompt: easynegative

Very Bad Image Negative

negative prompt: verybadimagenegative_v1.3

Combining with other Lora's should be possible but requires tinkering with the Lora strength value and can have varied results

Recommended prompt use this as a starting off point or if you want to generate similar results to me:

prompt: high res, best quality, 1girl, macro, giantess, city, full body, breasts, breasts apart, nipples, areolae, large breasts, nude, public nudity, exhibitionism, solo, uncensored

negative prompt: easynegative, verybadimagenegative_v1.3, low quality, low res, blurry, cropped, censored, censor_bar, safety panties, safety knickers, mosaic, watermark, artist name, username, signature, furry, bad anatomy, multiple breasts, extra limb, bad hands, missing fingers