Return to course: ComfyUI – Level 2 Stable Diffusion Art

ComfyUI - Level 2

LoRA

What is a LoRA?

How to use LoRA in ComfyUI

Adding 2 or more LoRAs in ComfyUI

Example: Use LoRAs to change the face and add style

Quiz - LoRA in ComfyUI

Textual inversion

What is textual inversion?

Using textual inversion in ComfyUI

Autocomplete for embeddings

Joining two strings in ComfyUI

Quiz - Textual inversion in ComfyUI

Image-to-image

Image-to-image workflow in ComfyUI

Multiple output images in image-to-image

Image variations with image-to-image workflow

Resizing an image before image-to-image

Quiz - Image-to-image in ComfyUI

Basic inpainting

Introduction - Basic inpainting in ComfyUI

Inpainting with a standard model

Inpainting with an inpainting model

Inpainting with an uploaded mask

Quiz - inpainting

Upscaling

Introduction - Upscaling in ComfyUI

Using an AI upscaler in ComfyUI

AI upscaler with a custom upscale factor in ComfyUI

Upscale with High Res Fix in ComfyUI

When to use High Res Fix

ControlNet - Part 1

Using ControlNet in ComfyUI

Copying human pose with ControlNet in ComfyUI

Copying lines and edges with ControlNet in ComfyUI

Copying depth information with ControlNet in ComfyUI

Multiple ControlNet in ComfyUI

Exercise - ControlNet in ComfyUI

Outpainting

Introduction to outpainting

Outpainting with SD 1.5 model

Outpainting with SDXL model

Outpainting with Flux Fill model

AnimateDiiff

Introduction - AnimateDiff (ComfyUI)

Setting up AnimateDiff in ComfyUI

AnimateDiff Prompt Travel

AnimateDiff video-to-video with prompt travel

Useful Tricks

Fast group muter

Aspect ratio preset

Style presets

Using style presets - SDXL

Using style presets - SD 1.5

Using style presets - Flux

Combining conditionings

Combining conditionings - Introduction (ComfyUI)

Prompt Break in ComfyUI (Conditioning Concat)

Conditioning Average and Combine (ComfyUI)

VAE

Quiz – LoRA in ComfyUI

You can use more than one LoRA on a checkpoint model. *

True

False

What does a LoRA model do? *

Speed up the sampling

Increase the image resolution

Reduce the VRAM usage

Add a new character or style to the checkpoint model

Which parts of a checkpoint model does a LoRA modify? *

Diffusion model

VAE

CLIP

KSampler

You can disable a LoRA node by: *

Deleting the node

Muting the node (Ctrl + M)

Setting the strengths of the LoRA to 0

Setting the number of sampling steps to 0

By-passing the node (Ctrl + B)