NSFW blue haired Nursity archive - Free
A downloadable asset pack
hello!!
Today, I would like to share with you the exciting journey of combining traditional pencil sketches with AI. In my early steps towards generating unique characters, I began with nine graphite pencil drawings. These initial sketches served as the foundation for the AI-generated models that followed.
To give you a glimpse of the remarkable transformation that took place, allow me to present the results of merging these hand-drawn sketches with AI.
----------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------------------
----------------------------------------------------------------------------------------------------------
With this collection of images, I started building a sufficiently large dataset to train it in Kohya. Both the dataset and the model will be made publicly available for anyone to use. The sale of generated images or the training of other SafeTensor files is allowed.
My LoRa in civitai: https://civitai.com/models/319032?modelVersionId=357751
----------------------------------------------------------------------------------------------------------
The following images are the content generated with my second refinement and can be found in the refinement v2 folder.
---------------------------------------------------------------------------------------------------------
Currently, there are 673 pristine NSFW images in the premium collection, including a zombie nurse, and 52 detailed and clean ones.
Clean dataset:
For cleaning:
----------------------------------------------------------------------------------------------------------
I have also added around 30 hospital backgrounds that I generated for my tests. All of them are in a resolution of 1380x2048. I won't put too much effort into them, but I do plan to fix some details in them.
Update: 3 new models:!
----------------------------------------------------------------------------------------------------------
I am updating this post to announce a significant increase in the number of images. Currently, there are two free files available. One is from the first iteration of the model in SDXL (a bit more realistic), and the second is the refinement of the models generated in Ponydiff (more focused on hentai/NSFW). Both Loras are available on my Civitai profile, and you can use them to generate as you like. Although v1.1 is much more consistent, you can use and download them as you wish.
All the paid content has been made free. However, as compensation for those who have supported me, I have opened a drive from which I will start working. From now on, I will work with this model. I consider my investigations with the styles of the first two refinements complete.
---------------------------------------------------------------------------------------------------------
content of my drive: (1.000 pics) pov, multiples views, multiples_angles
This content will be refined to achieve a style that I consider to be the final one. The work is slow, but the result is VERY good:
thank u very much *3*
muchas gracias por descargar/ayudar *3*
Download
Click download now to get access to the following files:
Development log
- v1.4May 29, 2024
- big update announceApr 20, 2024
- v1.3Apr 02, 2024
Comments
Log in with itch.io to leave a comment.
спасибо большое за такие красивые спрайты! я только учусь делать новеллы и буду использовать твои спрайты в своих играх! эта девочка просто милашка!
еще раз спасибо большое)))
Я очень рад, что вы можете использовать их в своем проекте, никто еще не отметил меня и не связался со мной, чтобы показать мне проект в работе, но я сам что-то делаю, чтобы иметь возможность показать спрайты в действии. Прямо сейчас я закончил работу над этим пакетом в его бесплатной модели и начинаю формировать новый пакет, который включает в себя больше, чем просто фотографии и, таким образом, сможет включать в себя работу над большим количеством фонов, значков и аксессуаров, ориентированных на визуальный роман.
Man... I wish there was a tutorial to do what you posted!
hooola, will you know how to use LoRas or train one? I have a guide in Civitai. I'll leave it here but in that guide I assume that you know how to train a LoRa.
Consistent Characters Lora Training // Entrenamiento de personaje consistente. | Civita
Thanks!
Guide to training your drawings in generative AI by Houdini (itch.io)
I have written a guide that took me hours. No one ever explained this to me, the knowledge I put in that guide is the result of a lonely path :c
Are we able to use these in a commercial project?
Yes, they are assets and you can use them without restriction on my part, you can modify them as well. If your project is moving forward, keep in mind that I have the tools to create specific assets or models to speed up the workflow.
Thank you very much, there is a lot of high quality content in these packs, I will use it in a small game, I am learning and I will use your pack for testing, thank you, not all heroes wear capes
I'm glad you can take advantage of it. I've been learning even more about using SafeTensors to achieve consistency. I will be rearranging these packages very soon
You, my friend, are using AI the right way.
You think? Thank you <3, I'm learning about her as much as I can. Years ago I did some background illustrations, this week I want to train a style and convert other drawings to the style of those illustrations.
Wow. That looks great. Nice combination of techniques. The files to download are the final resulting images? Or what is it?
Good work.
helloooou geru
let me explain
The rar files contain my created images, while the safetensor files are used to generate those images. The first safetensor is based on 9 graphite drawings and has been used throughout the process. The second safetensor is an improved version (v1.1), trained with more images and following the aesthetics of the "premium" package for better results.
The first safetensor has a 70% chance of generating the same nurse character. The second safetensor has an 85% chance of doing the same.
Regarding the rar files, the first one includes images created with different base models like albedo, dynavision, cheyenne, etc. They are not perfect but serve as a good base and can be used to create a dataset for this character. Some cleanup of backgrounds and fixing details is needed, but overall, they are not overly complex in terms of detail, with only a few errors.
The second rar file contains images generated in a specific anime style. It has more NSFW content, and I plan to clean it up. Currently, there are 55 images with transparent backgrounds, resized, and with fixed hands, eyes, and various details. I aim to increase this number to at least 110.
Oh, I see. It's pretty interesting. Thank you so much for your patience in explaining.