Create Magical Live AI Artwork with LCM and Stable Diffusion
Table of Contents
- Introducing Live AI Art Creation with LCM
- How to Set Up Live AI Art Creation
- Using the Canvas Tab for Live Input
- Alternative Methods for Live Input
- Upscaling Your Live AI Art Outputs
- Start Creating Your Own Magical Live AI Art Now!
Introducing Live AI Art Creation with LCM
Live LCM is the dream come true for AI artists. Recently there has been a lot of hype around tools like Korea AI that allow you to draw live and see AI art update automatically based on your inputs. However, while impressive, these tools tend to be limited in accessibility and flexibility.
That's why I've created a free, open-source alternative for live AI art creation using Stable Diffusion. With my workflow, you can use any models, any software, and any settings you like to create magical live-updating AI art.
Seeing the Hype around Live AI Art
You've probably seen examples online of platforms like Korea AI where a user can draw colors and shapes, while an AI model instantly generates detailed artwork based on those inputs. It's an extremely cool concept - you get to collaboratively create with an AI in real time. But up until now, these tools have been closed platforms limited to certain users.
Presenting a Free Alternative for Creating Live AI Art
I'm excited to introduce a free, customizable workflow for live AI art creation accessible to anyone. You can use it to experiment with different models, software, and settings to create your own magical live-updating AI art. And it's all open source!
How to Set Up Live AI Art Creation
Getting live AI art creation set up on your system is straightforward. There are just a couple of key files you need, which I'll link to here. Then it's a matter of installing them and configuring your Stable Diffusion setup.
Downloading Required Files and Models
First, you'll want to download the LCM Stable Diffusion model. This is a 5.14 GB file, but is necessary to enable the live art functionality. Place this file in your Stable Diffusion models folder. Next, download my custom Canvas Tab node. This allows you to draw input images live. Finally, grab the Efficiency Pack for best results.
Integrating Live Updating into Your Workflow
In Conda's U-I Manager, install any missing custom nodes that were just downloaded. Then load up the LCM model along with your preferred base model like Automatic1111's Dream Shaper. Make sure your Sampler is set to LCM and your Scheduler is set to Karas. Then in Extra Options, enable Auto Queue. Now when you generate images, it will automatically update based on any input changes!
Using the Canvas Tab for Live Input
The custom Canvas Tab node provides an easy way to draw images that will be used as input for live AI art generation. It's simple to use and allows you to bring your own creativity into the mix.
Painting and Seeing Live AI Art Update
Inside the Canvas tab, you have access to color palettes, brush sizes, layers, and more to draw whatever you want. As you draw, your creation will instantly be processed by Stable Diffusion to output AI generated artwork that incorporates your original input. It's magical seeing it evolve in real time!
Tips for Using the Canvas Tab
I highly recommend using multiple layers in your Canvas Tab drawings, keeping different elements like line art on separate layers from flat colors or shadows. This gives you more control. Also play around with CFG scale and denoising strength for best live update results.
Alternative Methods for Live Input
While convenient, the Canvas Tab in Comui has limited options. Luckily, I've configured additional ways to feed images into the live workflow allowing you to leverage external tools.
Loading Images from a Directory
You can actually save edit images externally using software like Affinity or Photoshop, save them to a designated folder as JPGs, and Comui will automatically load the latest version to use as input for live updates with Stable Diffusion. This allows more advanced editing.
Manually Loading Images for More Control
For ultimate control, you can also manually load input images. Just connect the Load Image node to your workflow, click choose file, and select any image you want. This bypasses the automated steps for a more hands on approach.
Upscaling Your Live AI Art Outputs
The live updated images from Stable Diffusion won't be very high resolution. But I've included an upscaling workflow to turn your AI art into gorgeous masterpieces. Just let it finish rendering and watch it enhance!
Getting High Resolution Versions
My bonus upscaling workflow will automatically run following live AI generation to increase output resolution drastically. This uses cutting edge tools like RealESRGAN to spit out crystal clear, high-res versions of the AI art you create.
Cropping to Desired Aspect Ratios
I even included options to automatically crop your upscaled outputs to different aspect ratios like square or portrait. No need to fuss with dimensions - just select what you want and enjoy frame-perfect AI art edits!
Start Creating Your Own Magical Live AI Art Now!
Thanks for reading about my live AI art creation workflow using Stable Diffusion. I'm excited to see what unique, mesmerizing art you make with it! Feel free to reach out if you have any other questions. Just remember - the possibilities are endless when you bring your creative ideas to life with AI.
FAQ
Q: What software and files do I need for live AI art?
A: You need Stable Diffusion, the LCM model, the Efficiency pack, and my custom workflow with nodes for live updating and upscaling.
Q: Can I create live AI art on my own computer?
A: Yes, by following the steps outlined in this post you can set up live AI art creation using Stable Diffusion on your own GPU.
Q: What is the benefit of the canvas tab for input?
A: The canvas tab allows you to quickly paint input images that will automatically update the AI art output in real-time.
Q: Why use alternative input methods instead of the canvas tab?
A: Other software like Affinity Photo provides more controls and tools for creating detailed input images to use.
Q: Do I have to do the upscaling part of the workflow?
A: No, upscaling is an optional final step to get a high resolution version of your live AI art.