11/3/2023 0 Comments Ema skye ace attorney deviantartIf you already have images to train with, upload them to your Google Drive's "lora_training/datasets/project_name" (old) or "Loras/project_name/dataset" (new) folder, and you may choose to skip step 2.Ģ️⃣ Scrape images from Gelbooru: In the case of anime, we will use the vast collection of available art to train our Lora. It will ask for permission, accept to continue the guide. Choose a simple name for your project, and a folder structure you like, then run the cell by clicking the floating play button to the left side. But fortunately, you can do most of this process automatically using my new □ dataset maker colab.ġ️⃣ Setup: This will connect to your Google Drive. You may gather your images online, and describe them manually. Thanks to this you can let those details be absorbed into an activation tag, which is a unique word or phrase that goes at the start of every text file, and which makes your Lora easy to prompt. This is because the Lora absorbs details that can't be described easily with words, such as faces and accessories. If there is something in all your images that you don't include in your tags, it will become part of your Lora. Let me describe how tags work in your dataset: You need to be detailed, as the Lora will reference what's going on by using the base model you use for training. For anime you'll need to use booru tags ( 1girl, blonde hair, full body, on chair, etc.). You can train a mediocre Lora with a bare minimum of 5 images, but I recommend 20 or more, and up to 1000.Īs for the descriptions, for general images you want short and detailed sentences such as " full body photograph of a woman with blonde hair sitting on a chair". For example a character may then be generated in new poses and in different clothes. As you add more variety, the concept will be better understood, allowing the AI to create new things that weren't in the training data. If all your images are face close-ups for example, your Lora will have a hard time generating full body shots (but it's still possible!), unless you add a couple examples of those. The quality of your dataset is essential: You want your images to have at least 2 examples of: poses, angles, backgrounds, clothes, etc. "1.png" and "1.txt"), and they all have something in common which you want the AI to learn. A dataset is (for us) a collection of images and their descriptions, where each pair has the same filename (eg. This is the longest and most important part of making a Lora. I won't cover them as I don't think they're very useful. LoHa, LoKR, (IA)^3: These use novel mathematical algorithms to process the training data. LoCon: Has more layers which learn more aspects of the training data. LoRA: The classic, works well for most cases. However there are also different types of Lora now: However I won't help you if you want to copy real people's faces without their consent.Īs you may know, a Lora can be trained and used for: This guide focuses on anime, but it also works for photorealism. We'll also be using a little bit of Google Drive storage. You can also pay $10 to get up to 50 extra hours, but you don't have to. We won't be doing that in this guide though, we'll be using Google Colab, which lets you borrow Google's powerful computers and graphics cards for free for a few hours a day (some say it's 20 hours a week). You could of course train a Lora in your own computer, granted that you have an Nvidia graphics card with 6 GB of VRAM or more. All while keeping some advanced settings you can use later on. Well, I've streamlined the process such that anyone can make their own Lora starting from nothing in under an hour. So many options and nobody explains what any of them do. Just try to read carefully, use critical thinking, and don't give up if you encounter errors. I'll try to explain these new concepts in an easy way. Knowledge about what Loras are and how to use them. You can even do this from your phone if you want to (as long as you can prevent the tab from closing). I just do it for fun.Īlso here's a page where I collect Hololive loras.Īn internet connection. However I do not support you if you want to use AI to trick people, scam people, or break the law. I believe all information should be free for everyone, including image generation software. I'm making this guide for the joy it brings me to share my hobbies and the work I put into them. If you're new to Stable Diffusion I also have a full guide to generate your own images and learn useful tools. In this guide we'll be using resources from my GitHub page. Click "Show more" below.Įven if you don't know where to start or don't have a powerful computer, I can guide you to making your first Lora and more! You don't need to download anything, this is a guide with online tools.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |