04 | Fake Everything
Advanced ComfyUI #
There are many extension for ComfyUI. One of the most powerful is the IP-Adapter. This extension allows us to extract semantic meaning from images, and use it to guide the generation process
Krita #
Krita is a professional FREE and open source painting program. It is made by artists that want to see affordable art tools for everyone. concept art, texture and matte painters, illustrations and comics. We can enhance Krita using the ComfyUI extension.
To set Krita up:
- Download and install Krita from here
- Download the ComfyUI extension from here
- Follow the instructions in the README file to install the extension
For the course, all the computers at K103 have Krita installed already, however we need to setup the plugin everytime we log into our accounts. To get things running:
- Open Krita
- Go to
Tools
>Scripts
>Import Python Plugin From File ...
- Find the
Krita-ai-diffusion.zip
file in theautomatic1111/krita-comfy
folder - Click on
Open
- Krita will ask you to restart the application
- Open ComfyUI and let it run in the background
- Once restarted, create a new canvas of any size, but ideally 1024x1024 pixels
- Go to
Settings
>Dockers
>AI Image Generation
- A new panel will be open.
- Click on
Configure
and selectLocal Server
- Press
Connect
Finally save your Krita file before starting the generation process. Now you can press the Play
and Record
buttons. The system will genenerate a new image every time you press Play
or if you are in the “Live mode”, everytime you make a change in the canvas. You can change the style on the dropdown menu, and the strength of the diffusion process with the slider.
Fake Music & Voices #
We can also use diffusion systems (or other AI implementations) to generate music and voices. Some of the tools that we can use are:
- Riffusion (Requires Login, limited generations, also works with Auto1111)
- Suno (Requires Login, limited generations)
- Udio (Requires Login, limited generations)
- ElevenLabs (Requires Login, limited generations)
- Tortoise TTS
- Resemble
- Voice AI (Requires Login, limited generations)
- Coqui TTS
- Coqui XTTS2
- Bark
- Bark Voice Clone
- OpenVoice
- Huggingface Voice Clone
- Huggingface Voice Cloning Space
- SadTalker
- Gooey LipSync
- Gooey TTS
- ComfyUI TTS
We can use the following prompt to record our own voice and generate a clone that we can use for further processing:
The tube was blown and the tire flat and useless.
Fruit flavors are used in fizz drinks.
The young kid jumped the rusty gate.
Pack the records in a neat thin case.
Let it burn, it gives us warmth and comfort.
The walled town was seized without a fight.
Next Tuesday we must vote.
The sofa cushion is red and of light weight.
Tack the strip of carpet to the worn floor.