TextualInversions

A commission of @findingavarice521.

Lyna Pérez (aka Lynaritaa) is an American model and social media celebrity with around 10M followers on Instagram.

This is a 1000-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 (lynaritaa:0.99), a woman, RAW, close portrait photo, long brown coat, ((turtleneck sweater:1.2), trousers, long haircut, slim body, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3 sharp focus, f 5.6"

34 year old German pornstar.

Still not 100% happy but this is my 6th version and I can't stand working on it any longer at the moment. I will try to improve it later on but I need a break from this ti as I was working 3 days straight on it.

_________________________________________________________________________________________

Avery, or averyofficial on Instagram, is an American Instagram model from LA with over 100k followers.

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 :)

Embedding of pornstar Alexis Fawx.

Standard weight

Clip Skip 1

7 CFG

DPM++ 2M KARRAS 30 steps

Upscale through hires fix or Controlnet 1.1 tiles

A commission of @Zorglub.

Rosario Dawson is an American actress of Cuban and Puerto Rican descent, known for her roles in many films, such as Sin City or Men in Black, and TV shows such as The Mandalorian.

This TI has been a tough one to train... I've had to repeat the process from scratch a couple of times, mainly because I couldn't get the eyes right. Reduced the dataset to 12 images, then removed some more pics and added a few new ones... I'm satisfied with the final results, but I may still update the embedding at a later point.

This is step 1325 of a TI trained on a dataset of 15 images. I used my standard 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 (r0sd4wn-1325:0.99), a woman, RAW, close portrait photo, long brown coat, ((turtleneck sweater:1.2), trousers, long haircut, slim body, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3 sharp focus, f 5.6"

Get better hands and place EMB in a negative prompt

获得更好的手部，将emb放置在负面prompt中

より良いハンドを獲得し、embを負のproptに置く

더 나은 손 얻기, 부정적인 prompt에 emb 배치

This embedding will allow you to force an image to be colored, black and white, or both. It can be called with just "b_w".

When using it in your positive prompt, think of it as pouring a big bottle of black and white onto your painting.

When using it as your negative prompt, think of that exact scene but in reverse.

It does sometimes refuse to work, but I find it a pretty handy tool so I decided to share it. ^.^

Emily Feld is an australian model with over a million followers on instagram (https://www.instagram.com/emily.feld/).
I have two very similar versions of it, here is the second one, working great for the face.
Tested with Liberty and Deliberate.

This Inversion of Vanna Bardot (Porn Thespian) was trained on 39 headshots, (a few flipped) at 2760 steps at a variable rate on v1.5-pruned, my preference is for a-ZovyaPhotoreal_Ultra but other checkpoints have decent results. Best placed early in prompt for accuracy.

If you like and care to help me pay off the 3080 chugging away before me you can kick some sand in my face here: https://www.buymeacoffee.com/admiralundr

An 8 vector embedding that emulates the style of a particular Japanese artist, trained on the NAI model.

You may notice in the image metadata different variations, like 'henreader-8' and 'henreader girl'. It's all the same embed renamed for various testing purposes, such as training the same dataset on different initialization tokens and at different sizes; the presented embedding was the best result.

19 year old American Only Fans model.

Embedding of pornstar Nikki Benz. 99% chance you already nutted to her, if not consciously then probably in your sleep.

Standard weight

Clip Skip 1

7 CFG

DPM++ 2M KARRAS 30 steps

Upscale through hires fix or Controlnet 1.1 tiles

7.5k steps on 57 images