Let's have fun with LoRA! How to install and use the training file! [Stable Diffusion]
🐸This video explains how to install and use trained files created with LoRA, the learning function of the image generation AI "Stable Diffusion webUI AUTOMATIC1111" (local version). It is based on specifications as of March 2011.
🐸This video focuses on the "Local Version AUTOMATIC1111," not the so-called "Colab Version," "One-Touch Version," or "NMKD Version."
🐸This video was created for technical research purposes. It also provides an example of how to use the software and is not intended as a recommendation for its use.
🟢What is LoRA?
It is a learning method. Generally, it refers to the training files generated using this method.
🟢What happens when you install it?
By applying LoRA, you can create pretty much anything.
For example, erotic images of copyrighted characters, or illustrations in specific styles like ukiyo-e and sumi-e.
It can also create so-called "eye-collage" (eye-collage) effects, so use at your own risk.
---------------------------
What is Stable Diffusion webUI AUTOMATIC1111? If you're wondering, click here:
🟠How to install Stable Diffusion webUI AUTOMATIC1111
• 日本一わかりやすいStableDiffusion WebUI AUTOMATIC11...
↑(As long as your computer meets the requirements) Follow the instructions in this video to install it.
🟠Author AUTOMATIC1111's page
https://github.com/AUTOMATIC1111/stab...
↑Please be sure to read this before installing WebUI.
---------------------------
🐸Video Supplement
🐸About Creating Thumbnails (from 03:10)
If you right-click and save the thumbnail on Meenow's page, it will become a jfif file. However, jfif files are not recognized by 1111. Therefore, we recommend ① right-clicking the thumbnail, ② copying the image, ③ pasting it into Paint, and ④ saving it as a jpg.
*If your computer freezes when clicking the Hanafuda icon due to computer power limitations, please delete the thumbnail.
🐸About Using Multiple LoRAs (from 06:59)
While it's not a set rule, when using multiple LoRAs simultaneously, it's recommended that the total weight not exceed 1. For two, use 0.5*2, for three, 0.3*3, etc. A total weight of 1.5 will result in a jagged image.
✅LoRA used
🟢Tifa_meenow (Meenow)
https://civitai.com/models/11367/tifa...
↑※As of March 27th, this is 404 (´･ω:;.:…
A LoRA based on Tifa from FF7 Remake and a Chinese model.
It looks a bit like a Chinese Tifa.
Trigger words: tifa lockhart, red_eyes, black hair, long hair,
🟢Cute Joints Dolls Lora (set999)
https://civitai.com/models/11204/cute...
↑As the name suggests, this LoRA turns into a doll.
Without the trigger word, the doll will have no joints. Optional.
Trigger words: doll, dd,
※Although not mentioned here, careful creators may include a backslant (\) in their LoRA trigger words.
Example: Satono Diamond \(umamusume\)
The backslant is used to recognize the following parentheses as "normal parentheses, not emphasis syntax." This is to avoid confusion with the character with the same name, LoRa. Therefore, if you are using it as a normal trigger word for a single character, you can ignore it and just enter "Satono Diamond" and it will be recognized as a trigger in most cases.
✅Model and VAE used in the video
AnythingV4 (Anything V4) by andite
https://huggingface.co/andite/anythin...
GuoFeng3 (GuoFeng3) by xiaolxl
https://huggingface.co/xiaolxl/GuoFeng3
↑A beautiful Chinese live-action model
LiveArcaMix (Live Arca Mix) by 0RisingStar0
https://huggingface.co/0RisingStar0/L...
840000VAE
https://huggingface.co/stabilityai/sd...
✅This is the chillout mix mentioned briefly in the video.
ChillOutMix
https://civitai.com/models/6424/chill...
↑Login Required
※As you can see from the page, this model has complicated rights issues, so use it at your own risk. This channel does not handle it.
🐸🐸🐸Administration space below🐸🐸🐸
Timestamp
00:00 Introduction
00:40 What is LoRA?
01:26 Get the LoRA file!
03:37 How to use LoRA
05:33 Unlocking LoRA's capabilities
06:59 How to use multiple LoRAs (Tifa LoRA + Doll LoRA example)
BGM used: "Unleash", "Girlish", "Sugary Pop", "OPENESS" - all written by SHUNTA
#generativeai #AIillustration #StableDiffusion #StableDiffusiontutorial #generativeai #AIart