Creepy, Cute, & Magical Impressions
I have been shooting for a full SDXL (and now Illustrious) Creepy Cute Magical and I think I did OK but it's not nearly as capable as the Flux version. It seems to have infused the Creepy, Cute, and Magic but also a bit of other. So rather than pile in a different version to the main I decided to roll this out as a Creepy Cute Magical Impressions version. It has the colors, it has the flair, it has a lot of it's own character. There are some things that I think have come out even better than with Flux (I am a fan of impressionism as my other LoRAs can attest after all) but also some oddities.
Since it's SDXL or Illustrious you may need to run more than one run to get to what you want. But it's definitely worth using and you may find some gems that weren't what you thought you wanted. It will combine well with other SDXL or Illustrious LoRAs. I hope you like it and have fun with it!
Fair warning... I have found the output to be consistent in quality and color but image differences between different checkpoints seems to be immense. It worked well with all that I tried (and with Illustrious I tried a bunch as the examples will show) but your output may vary.
Enjoy!
A playful fusion of whimsy and enchantment, CreepyCuteMagical is all about adding a vibrant, otherworldly twist to your art. In this style, you’ll find characters with big, expressive eyes; luminous, high-contrast colors; and an overall atmosphere that balances “cute” with just the right touch of “creepy.” Think sweetly sinister creatures, magical beings with hidden powers, or haunted carousels that practically glow with eerie charm. This LoRA is designed for 2D or 2.5D illustrations, giving a painterly feel that pops right off the screen.
How to Use
Keyword:
CreepyCuteMagicalRecommended Weights: Ranging from 0.5 to 1.5.
At lower strengths (0.5–0.7), your original prompt style has more influence, while subtle magical-cute elements sneak in.
Around 0.8 is a sweet spot for a balance of vibrant, whimsical flair for SDXL and 1.0 for Illustrious.
At higher strengths (1.0–1.5), expect a stronger push toward abstract and deeply impressionist forms, vivid color shifts, and more abstract effects.
This LoRA works beautifully on a wide range of subjects: animals, people, detailed architecture, fantastical landscapes—anything you can dream up. The style does well with complex prompts where it adds a layer of dreamy, fun energy.
Behind the Scenes
I created CreepyCuteMagical Impressions using the core training data I used with the Flux version adapted to SDXL, with each bit of training data brimming with intricate designs and a dash of the surreal. You’ll see these traits popping up across the sample images:
Bold Color Palettes – Vivid pinks, blues, and unexpected neons that bring a whimsical feel.
Expressive Characters – Large, sparkling eyes and gentle (or spooky!) facial expressions that add personality.
Magical Elements – Glowing orbs, floating shapes, and detailed accessories for that enchanted vibe.
Fantastical Creatures & Scenes – Whether it’s a ghostly companion or a skeletal carousel, there’s always a bit of mystery sprinkled in.
Because the LoRA stems from many different sources of inspiration, it doesn’t adhere to just one singular art style. Instead, it’s a melting pot of all the qualities I love: lively colors, delicate details, and an undercurrent of magic.
Keep me motivated, share the whimsy
Give this LoRA a spin, experiment with prompts, and see what hauntingly adorable masterpieces you can create. Be sure to:
Give a thumbs up if you enjoy using it.
Post your first CreepyCuteMagical Impressions image to share the buzz! Your creations keep me inspired and help this LoRA evolve.
Let’s make some whimsical, spine-tingling magic together!
Description
Several runs and tweaks this is the first I feel represents the style with the speed and ease of SDXL
FAQ
Comments (7)
This looks incredible! I love it!
Thanks so much. I hope it works great for you!!
This is some beautiful work. What was your criteria for images selection for this one. I am studying lora training a bit and any information is valuable 🙂
I actually first did several rounds of training with Flux towards this model -Creepy Cute Magical - v2.0 Flux ER | Flux LoRA | Civitai and then I used a set of the best images from that with consistent themes. Things like coloring, tones, eyes, body shapes, consistent "feel". Then I ran a bunch of epochs against those and tested them to find the (highly subjective) best one to share.
@Daalis Thank you. This is helping a lot.😊
@chris4wards413 Glad to hear it. If I can answer any other questions feel free to ask.
@Daalis 🙂🤗👍
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.















