Train ComfyUI Lora On Your Own Art!
ฝัง
- เผยแพร่เมื่อ 6 ก.ย. 2024
- Whether you're a comic artist, CG artist, or looking to infuse unique styles into your projects, this tutorial will elevate you above the pack of AI artists! Use your own art skills to train Loras and take control of ComfyUI or Automatic 1111 !
This tutorial is not just about learning to use a tool; it's about unlocking new possibilities in your artistic endeavors. Perfect for graphic novel creators or anyone looking to bring their art, sketches, and designs to life with precision and flair.
Click the links below to access the workflows and start empowering your creative process today!
~Auto train Advanced
github.com/hug...
~Download the workflows~
Simple Lora Test
github.com/cre...
Lora with Pose
github.com/cre...
~Models~
sd_xl_base_1.0.safetensors
your custom lora
OpenPoseXL2.safetensors
Yolox_l.onnx
dw-ll_ucoco_384.onnx
~Nodes~
CheckpointLoaderSimple
LoraLoader
LoadImage
PreviewImage
SaveImage
Prompts Everywhere
Anything Everywhere
CLIPTextEncode
KSampler
EmptyLatentImage
Get image size
VAEDecode
ControlNetApplyAdvanced
DWPreprocessor
I noticed other lora trainings involve giving it captions. But your workflow didnt mention that. What are the drawbacks of not including captions when training a lora?
2:18 i dont see any options that say runwayml 1-5, or any 1-5 options for that matter under the base model dropdown. Why is that? and how do i fix that?
i cant get pulid / faceID / instant id to recognize mascot faces. if i train a lora, then insert that lora into the workflow, can i get my prompt to draw the mascot as it should look in real life?
where I can find Get image size node?
How to install localhost?
Install Comfyui locally?? Download from its git page the self contained package for Windows. Or are you talking about something else?
What's the hardware requirements
For training - computer with a web browser that can surf the internet :) The training happens for free on Colab, with their TPUs.
In the video a RTX 2060 is being used with ComfyUI... what do you think? Tech spec fits your set up needs?
Using AI for Voice... lol
Most humans left Earth a while ago… AI 🤖 lets us stream content to TH-cam from Mars to those who can’t go off world, you know, like Blade Runner, just not as exciting.