Hey guys!
We fine-tuned a dreambooth model on top of hentai model , the training images are some anime images generated using some anime-generation app.
Example Prompts
positive prompt = "model photoshoot of zwx, masterpiece, best quality, ultradetailled, natural, Naked, anime, NSWF, No-Clothes, Giantess, fully-Naked, from above, Sakura Haruno from the Naruto series, pink hair, she has the most big and huge naked boob in the world and beautifull natural tits, milk herself"
negative prompt = "(deformed iris), (deformed pupils), (((cropped))), out of frame, worst quality, low quality, (ugly), (duplicate), double, (mutilated), (extra fingers), mutated hands, poorly drawn hands, poorly drawn face, (mutation), deformed, blurry, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, disfigured, oversaturated, low-res, mangled, old, surreal, calligraphy, sign, writing, watermark, text, body out of frame, bad anatomy"
Join our Discord for any questions or feedback!
Settings
Resolution = 512 x 512
Sample images were generated with CFG = 7 and inference steps =100.
Sampler is recommended using Euler a
Description
FAQ
Comments (5)
Can't help noticing that 3 out of 4 examples have cropped head/head out of frame. Is this a typical problem for your dataset?
the training dataset is not cropped, the code has some croping mechanism. worth to change the data to 512*512 before training
Cropped images and no prompts?
sorry guys, just added the prompt. Join our Discord for any questions or feedback!
imo looks like overtuned downgrade
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.







