Главная
Study mode:
on
1
Intro
2
Tools you'll need
3
Dataset introduction
4
Reasoning for using a character LoHa
5
Captioning the character
6
Character LoHa training settings
7
Initial test of character LoHa
8
Choosing the best epoch
9
Verifying the chosen epoch LoHa integrity
10
Mini veadotube mini overview
11
Choosing the style LoRA
12
Check the raw style LoRA
13
Prompting with both character LoHa and style LoRA
14
Technique to vary seed within a radius
15
Compositing bust shots together
16
Overview of base full body image
17
Outpainting the full body
18
Outpainting the bust shot
19
Upscale the body
20
Upscale the bust
21
Separate head from background
22
Preparing the body for bust photobash
23
Rasterizing a smart object
24
Liquifying for better seams
25
Refining the entire character
26
Preparing head for smart object simplifies character expression replacement
27
Creating the open mouth expression
28
Creating the closed mouth expression
29
Creating the closed eyes expression
30
Making the closed eyes a global toggle
31
Removing the background from the entire character
32
Exporting in the correct format for veadotube mini
33
Setting up pngtuber in veadotube mini
34
Using veadotube mini with OBS
35
Closing thoughts
Description:
Learn how to create a PNGtuber character using Stable Diffusion and Character LoRA in this comprehensive tutorial. Explore the process of dataset preparation, character captioning, and LoHa training settings. Master techniques for testing and selecting the best epoch, integrating style LoRAs, and prompting effectively. Dive into advanced methods like seed variation, bust shot composition, and full-body outpainting. Discover the intricacies of upscaling, character refinement, and expression creation. Finally, gain insights into background removal, proper export formatting, and setting up your PNGtuber in Veadotube mini for use with OBS.

Stable Diffusion - Create a PNGtuber with Character LoRA

kasukanra
Add to list
0:00 / 0:00