Cat bikini. All images are generated with A1111 on PC. Trained with Kohya with 50 pics and a realistic checkpoint and tested with my Illustrious model available on TA/HUB.

This model does not use any real-life or celebrity content. Please do not use this model for any real-life, celebrity, or pornographic content.

Illustrious Weight 0.7 - 0.8 CFG 5 on realistic checkpoints.

IMPORTANT: All my models work only if the trigger words and the recommended negative and positive prompts and embeddings are used. If you have a different result from the images, it is because you have not used the recommended settings. Too many loras and/or embeddings can ruin the model's result. To have images like the ones I made, use ONLY the recommended settings. If you use the indicated settings (below), you should be able to create images like the ones shown in the photos/videos. The delicate images with an ethereal gradient are intentional and included in the lora. With anime models, it could give completely different results. In the images, I used my semi-realistic model available on TA..

Example positive prompt:

a beautiful girl wear a colors catbikini outfit, cat ears, g-string, black thighhighs, printed bikini, large breasts, g-string thong, highleg, cleavage, big breasts, sunny day, slender body, big thighs, large hips, skin pores, soft curves, mole, eyelashes, happy, look at viewer, [hairstyles, haircolors, pose and location]

You can change colors of dress.

For behind views, you can add "from behind, ass, round butt, looking back" and delete "cleavage"

Settings For Illustrious version:

IMPORTANT, PLEASE READ : If you get different results, the prompt was probably not followed and an unsuitable checkpoint was used. All my LoRAs have been trained on realistic models. The videos I post represent what you can achieve; even though the video is animated, it reflects the quality and what the LoRA can do. The first frame of the video is the image I created using the LoRA and then animated, using the parameters I recommended. To achieve a realistic result, a realistic ILLustrious non-merged checkpoint should be used (or Semirealistic based on realistic checkpoint) . If you use merged models with anime models, the result may be very different from mine. Do not use corrective loras in addition to the main lora because they often only ruin the result. If possible, do not exceed 150 tokens in prompts. Too many words in prompts do not improve the result. The parameters I use are the following :

Steps: 35, Sampler: Euler a, Schedule type: Automatic, CFG scale: 4.5, Size: 512x768, Denoising strength: 0.35, ADetailer model: face_, ADetailer confidence: 0.5, ADetailer dilate erode: 4, ADetailer mask blur: 8, ADetailer denoising strength: 0.25, ADetailer inpaint only masked: True, ADetailer inpaint padding: 32, ADetailer use separate CFG scale: True, ADetailer CFG scale: 4.5

For Adetailer face, you can choose the better: face_yolov10x_

Reposting on other sites is prohibited.

It is strictly prohibited to distribute outside TA-HUB this model or hosting services in third-party image generation services, or sell my images on other site. You can use this model for personal use on HUB or A1111 on your pc. Not for sell or repost on pixai or other sites.

If you appreciate the lora, please consider leaving us a review and sharing the images generated using this model. Your feedback is valuable to both us and the AI community. Thanks.

Cat bikini - v1