Trang phục Nhật Bình phụ nữ Huế, Việt Nam

A model trained off of James Cabello "artwork" (https://twitter.com/JamesCabello_) Does not compare to the original, but does have a unique look.

Trained on NovelAI (clone) for 10k steps off of 61 images. Compatible with other TI's and LoRAs. I recommend a weight of 1.1 or less. Works on all models, ironically worst results with the model it was trained on.

Clip skip can be 1 or 2, but who is running without it? Actually running without clip skip seems to always remove the text and occasionally get better results. Text might also get canceled out depending on other TIs and LoRAs used.

I couldn't find a way to get a model to do this natively so I created some sample images using the Art of MTG embedding. It's a model that knows what a dryad is. I made a subject textual inversion which is now capable of making green skin and leaf & vine textures. Add a pattern overlay to smooth skin in order to coax growth patterns.

bad_prompt Version 1 and 2 (reupload from Huggingface)

Foreword

Since alot of people asked for it. Here is a reupload to civitai for my negative Embedding.

Disclaimer: A new negative embedding is still under develop and I will upload it on civit soon!

~Text from Huggingface~

Link: https://huggingface.co/datasets/Nerfgun3/bad_prompt

Idea

The idea behind this embedding was to somehow train the negative prompt as an embedding, thus unifying the basis of the negative prompt into one word or embedding.

Side note: Embedding has proven to be very helpful for the generation of hands! :)

Usage

To use this embedding you have to download the file aswell as drop it into the "\stable-diffusion-webui\embeddings" folder.

To activate the embedding, you will have to write the name of the file in the negative prompt!

Please put the embedding in the negative prompt to get the right results!

For special negative tags such as "malformed sword", you still need to add them yourself. The negative embedding is trained on a basic skeleton for the negative prompt, which should provide a high-resolution image as a result.

Version 1:

Issue: Changing the style to much.

~not downloadable on civit~

Version 2:

With this version I tried to reduce the amount of vectors used, aswell as the issue with the changing artstyle. The newer version is still a work in progress, but its already way better than the first version. Its in files section!

To use it in the negative prompt: "bad_prompt_version2"

I hope you enjoy the embedding. If you have any questions, you can ask me anything via Discord: "Nerfgun3#7508"

This is a textual inversion trained on pictures of wrestler Bianca Belair, aka Bianca Blair Crawford. This ti was requested by user Rin_chan.

I hope you enjoy using this ti as much as I do.

You need to play a bit with the prompt for her hair to come out right. I couldn't always get it right tho.

Still learning and improving the progress of the training, any feedback is welcome.

Feel free to leave me a tip if you like what I am doing :)

42 year old American actress.

Sara Jean Underwood is an American model, actress and TV host. She was the Playmate of the Month of July 2006, and Playmate of the Year in 2007.

1500-step TI trained on a dataset of 18 images with my usual settings.

Appreciate my work? My TIs are free, but you can always buy me a coffee. :)

Curious about my work process? I have summarized it here.

How to create a good prompt using my TIs

You're obviously free to experiment, but bear in mind that my TIs are trained with a more or less fixed phrasing, that normally starts with:

"photo of EMBEDDING_NAME, a woman"

So I recommend always starting your prompt like that and then building the rest of the prompt from there. For instance, "photo of (s4runderwood:0.99), a woman, RAW, close portrait photo, long brown coat, long haircut, pale skin, slim body, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3 sharp focus, f 5.6"

Sunny Leone

Sunny is a Canadian-Indian-American former pornographic actress now working as an actress and model in the Indian and American film industries. So I guess it would be correct to say that she left us too soon ;)

This is a textual inversion, if you want a LoRA of Sunny look HERE

Like my work? Buy me a beer!

[ SDv1.4, 4e-03, 4400, v/s .202 ]

A commission of @Zorglub.

Mekia Cox is an American actress and dancer, known for her roles in series such as 90210 or The Rookie.

This is a 500-step TI trained on a dataset of 18 images with my usual settings.

For some reason, it seems like reducing the token count to 2 is making my TIs train faster now... I'm not sure, it might be just a lucky streak, but both this TI and Sara Underwood's looked almost ready from a very early stage in their respective trainings. I'll keep testing with this setting.

Appreciate my work? My TIs are free, but you can always buy me a coffee. :)

Curious about my work process? I have summarized it here.

How to create a good prompt using my TIs

You're obviously free to experiment, but bear in mind that my TIs are trained with a more or less fixed phrasing, that normally starts with:

"photo of EMBEDDING_NAME, a woman"

So I recommend always starting your prompt like that and then building the rest of the prompt from there. For instance, "photo of (m3kiac0x-500:0.99), a woman, RAW, close portrait photo, long brown coat, long haircut, pale skin, slim body, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3 sharp focus, f 5.6"

This is a textual inversion trained on pictures of wrestler Becky Lynch, aka Rebecca Quin. This ti was requested by user Rin_chan.

I hope you enjoy using this ti as much as I do.

Still learning and improving the progress of the training, any feedback is welcome.

Feel free to leave me a tip if you like what I am doing :)

This is a textual inversion trained on pictures of wrestler Tay Melo, Tay Conti aka Taynara Melo Guevara. This ti was requested by user rimale.

I hope you enjoy using this ti as much as I do.

Still learning and improving the progress of the training, any feedback is welcome.

Feel free to leave me a tip if you like what I am doing :)

24 year old Indonesian nude model.

This is a textual inversion trained on pictures of wrestler Iyo Sky, aka Masami Odate.

Make sure to give enough weight to her clothing, it seems it ti likes to depict her in her wrestling attire.

Still learning and improving the progress of the training, any feedback is welcome.

Feel free to leave me a tip if you like what I am doing :)

I noticed that on some models there is a problem that would depict a girl in any clothes, but without panties, with a naked bottom. You need to enter a lot of tokens, and play with their weights. This textual inversion was created to simplify this task.
__

Я заметил что на некоторых моделях существует проблема в том, что бы изобразить девушку в какой либо одежде, но без трусов, с голым низом. Необходимо ввести очень много токенов, и играть с их весами. Данная текстовая инверсия создана что бы упростить эту задачу.

A commission of @Springbok707.

Kim Basinger is an American actress and model, winner of an Academy Award and known for roles in many notable films, including Batman and L.A. Confidential. My goal with this embedding was to somehow capture her looks of the late 70s and 80s.

This is a 1200-step TI trained on a dataset of 18 images (partly provided by @Springbok707) with my usual settings. However, this time I've tried a couple of new things: I left the initialization text completely blank (usually I leave a "*" symbol instead) and reduced the token count to just 2 (it's usually 5 for me). I think the results have been very good, so I may keep these settings in the future.

Appreciate my work? My TIs are free, but you can always buy me a coffee. :)

Curious about my work process? I have summarized it here.

How to create a good prompt using my TIs

You're obviously free to experiment, but bear in mind that my TIs are trained with a more or less fixed phrasing, that normally starts with:

"photo of EMBEDDING_NAME, a woman"

So I recommend always starting your prompt like that and then building the rest of the prompt from there. For instance, "photo of (youngk1m:0.99), a woman, RAW, close portrait photo, long brown coat, long haircut, pale skin, slim body, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3 sharp focus, f 5.6"

Sofía Margarita Vergara Vergara (born July 10, 1972) is a Colombian and American actress and model.

This TI is intended to get her looks when she filmed Four Brothers.

You can invite me to a Ko-fi if you want!

v1: This inversion of Kaylin Slevin (Instagram Bimbo) was trained on 18 images (including flipped ones) at 2600 steps on v1.5-pruned and pairs very well with Fake Lips: https://civitai.com/models/18264 for a bit of 'authenticity'. I got best results with s1dlxbrew: https://civitai.com/models/4717/ as a checkpoint.

This is my first upload, feel free to be cruel. v1 had the vector tokens set to 8 so you may need to (full body:1.6) or more to get past south of her navel in your pics. Best results at 768x768. Best placed early in prompt for accuracy. (note in the second 768x1024 pic her iris' are InPainted blue)

v2: Trained at 2550 steps at a variable rate on v1.5-pruned, works more consistently across more checkpoints but I still got best results with s1dlxbrew. Same dataset as V1 but upscaled and refined more carefully with less vector tokens making her more flexible when using text2img.

This is my sixth publicly released Textual Inversion, called Style-Swampmagic. This model was trained on a single image using DreamArtist. This TI gives things as the name implies, a swampy/earthy feel. Things come out extremely mossy with foliage anything that you can imagine when you think of swamps!

The strength of the TI output depends on the prompt used and can range from subtle to overwhelming. It's important to adjust the weights or where you put the TI in the order of the prompt.

All display images have been upscaled, so the seeds/parameters for 1:1 image replication will not be correct. I am working on adding the original prompts/seeds to the description of each image. However, even with this information, it may be challenging to replicate an image exactly 1:1. You can refer to this comment for further explanation: https://civitai.com/models/14065/faetastic?commentId=69125&modal=commentThread

If you're interested in upscaling images to achieve high-quality results, I've written a guide that you may find helpful: https://ko-fi.com/post/The-Double-Headed-Issue-E1E0JO2UV

A request by @Zorglub. Melanie Scrofano is a Canadian actress, having starred in series such as Wynonna Earp, Bad Blood and more recently in Star Trek: Strange New Worlds

The embed has been trained on SD 1.5 and works well on most 1.5 based models. Occassionally on certain models birthmarks can appear on her neck. Negative prompting largely prevents this, but inpainting always solves the issue.

If you enjoy my creations or have a face you'd like to commission, consider supporting me on Ko-Fi :)