By H.W.
A happy hedgehog resting on a fluffy rug
EXTRA! EXTRA! The Daily Hedge is Showcasing Real Art!!
That's right! Our founder, Arthur P. Bolderhedge, has embarked on a new venture: artwork by actual humans! Real, authentic art! This venerable newspaper has begun its Handmade Hedgehogs series! These hand drawn, hand painted, and sometimes even hand sculpted hedgehogs are sure to entertain and delight!
WOULD YOU LIKE YOUR ART TO BE SHOWN HERE? I know I would! If you do too, send a JPG, PNG, or similar image to our founder's email: thedailyhedge@outlook.com. Include your initials to be properly credited!
WHAT IS THIS?
Typically, this is a site that generates a new A.I. hedgehog picture every morning. I use an A.I. system called Stable Diffusion, and the image prompts are randomized. The text of the prompts is mostly sourced from me and my friends. However, right now we're running human made hedgehogs!
WHY IS THIS?
Like all right-thinking people, my mom likes hedgehogs. I made this site in case she's ever feeling gloomy because sometimes, when you're feeling down, a nice picture of a hedgehog can brighten your day. So if you're having a real time of it, I hope these pictures cheer you up, because even people who aren't my mom deserve to smile (sometimes).
CREDITS
This website could not exist without...
Black Forest Lab's Flux: The core technology, an open source generative A.I. system named Flux.
ComfyUI: The flexible, scriptable program that generates images.
ComfyScript: A Python front-end to ComfyUI so that I don't have to load weird JSON files.
Flask: A Python website framework that I learned so I could make this site. Hope I didn't mess it up too badly.
DynamicPrompts: A Python module that strings together random image prompts.
BareBones.CSS: A basic responsive CSS that works well and isn't as complicated as Bootstrap.
Fooocus Project: I've included general style prompts from TWRI, 3Diva, and K3nt3L
My Friends & Family: They let me spend time on this, and helped come up with some really great prompts and ideas.
MODELS I USED
The latest workflow uses a variety of models.
Chroma v27 Image generation model used in the latest workflow.
LLaVA 1.6 Mistral 7B A large language model that tweaks the daily prompt.
T5XXL FP8 E4M3fN Scaled A CLIP model that relates the prompt's ideas to a potential image.
Real-ESRGAN 4X Plus A good general purpose upscaling model. Used in the final stage of image generation.
ZavyChromaXL A SDXL image model that's used to polish the image during the upscaling step.
Dreamshaper XL Turbo Another image generation model. Images before February 14, 2024 use this model.
The workflow that generates the images is available here. The previous version (used from February 16, 2024 to June 11, 2025) is here, and the original version (pre Februrary 16, 2024) is here.