Stability AI just released an new SD-XL Inpainting 0.1 model. Here is how to use it with ComfyUI.

Stable Diffusion XL has trouble producing accurately proportioned faces when they are too small. Today, I learn to use the FaceDetailer and Detailer (SEGS) nodes in the ComfyUI-Impact-Pack to fix small, ugly faces.

I previously tried Thibauld’s SDXL-controlnet: OpenPose (v2) ControlNet in ComfyUI with poses either downloaded from OpenPoses.com or created with OpenPose Editor. Here are a few more options for anyone looking to create custom poses.

Yesterday, I tried out Stability AI’s four Control-LoRAs but mentioned that I did not understand the output of the Revision “image-mixing” workflow. I’ve since done a bit more experimentation...

Less than a week after my post testing diffusers/controlnet-canny-sdxl-1.0, along comes Stability AI’s own ControlNets, which they call Control-LoRAs! Not one but 4 of them - Canny, Depth, Recolor and Sketch models!

As promised in my last post, today I am testing out Thibaud Zamora’s SDXL-controlnet: OpenPose (v2) model using ComfyUI. I keep saying I’ll keep my posts short but never do...

Recently I tried Fooocus by Lyumin Zhang (Illyasviel) which fulfills its promise to allow one to “Focus on prompting and generating” - it is certainly easy to use! But shortly after its release, someone has “ported” the code to ComfyUI as a Custom Node! So of course it’s time to test it out...

The Stability AI documentation now has a pipeline supporting ControlNets with Stable Diffusion XL! Time to try it out with ComfyUI for Windows.

In this post, I experiment with latent scaling and latent compositing with SDXL 1.0 using ComfyUI. That is to say, increasing / decreasing the size of the image, and combining multiple images into one à la green screen (chroma key) compositing.

SDXL 1.0 uses two text prompts used to guide image generation. In my first post, SDXL 1.0 with ComfyUI, I referred to the second text prompt as a “style” but I wonder if I am correct. I have no idea! So let’s test out both prompts...

Onward with SDXL and ComfyUI! Sometimes I want to tweak generated images by replacing selected parts that don’t look good while retaining the rest of the image that does look good. Rather than manually creating a mask, I’d like to leverage CLIPSeg to generate a masks from a text prompt.

Yesterday I mentioned in passing that my Nvidia RTX 2060 with 12GB could not run both SDXL 1.0 Base and Refiner models in a single ComfyUI workflow. Today, I show you my workaround and also experiment with adding the SDXL 1.0 Official Offset Example LoRA to the workflow.

Stability.ai has released Stable Diffusion XL (SDXL) 1.0 (26 July 2023)! Time to test it out using a no-code GUI called ComfyUI!

Not long ago, I was reminded of one of my favourite digital artists, Linda Bergkvist a.k.a. Enayla. I love her macabre yet stunningly beautiful and evocative art, but alas, she’s not posted her art since around 2007. This post is a departure from my usual techie stuff -- it’s just appreciation post!

Epic Diffusion recently came to my attention, a high-quality merge of various models by John Slegers: “Epîc Diffusion is a general purpose model based on Stable Diffusion 1.x intended to replace the official SD releases as your default model. It is focused on providing high quality output in a wide range of different styles...” Figured I’d give it a spin.

Recently (around 14 December 2022), Apple’s Machine Learning Research team published “Stable Diffusion with Core ML on Apple Silicon” with Python and Swift source code optimized for Apple Silicon (M1/M2) on Github apple/ml-stable-diffusion. Here I’m trying it out on a MacBook (though the code also works on iPhones and iPads)...

I refactored my previous Stable Diffusion code, to clean up, OO it a little, and add new features like tiling, upscaling, PNG metadata. As I mentioned before, I don’t understand AI/ML... but I do understand programming! So here is my new, more elegant Simple-SD v1.0 Python script.

I have more ideas for Stable Diffusion. My nights and weekends are consumed! This time: For inpainting, why create a mask image manually, when A.I. can automatically build a mask from a text prompt? Someone much smarter has already published a paper (arXiv:2112.10003 [cs.CV]), with source code, to do just this!

More Stable Diffusion! This time attempting to add inpainting / masking based on my previous code, to merge both txt2img.py and img2img.py capabilities, disregarding the out-of-box inpainting.py code, which does not have parameters for positive or negative prompts. Keyword being attempting...

I’ve been playing around with the Stable Diffusion scripts a little (to be exact, Ben Firshman’s version). To help me understand the script, I decided to re-write it the way I prefer to use it... either breaking or optimizing it in the process :P