
触发词
文件清单
下载后放进对应的 ComfyUI 目录。models/ 指 ComfyUI 的模型根目录。
说明
Storybook Spark : Umbrium Painterly Cartoon Style
作者:umbrium2835
Painted Character gives your images a colourful, dimensional cartoon look, combining expressive faces and exaggerated shapes with soft painted shading, glossy highlights and selective outlines. Hair, clothing and facial features take on a sculpted, illustrated appearance, with strong lighting helping give the characters depth. The style sits between cartoon character art and polished fantasy illustration. It can bring a more animated look to portraits, full-body characters and scenes with multiple people, while retaining painted detail in clothing and surroundings. How to use
- Add linestyle to your prompt.
- Describe your subject, clothing, pose, expression and setting normally.
- Start at strength 1.0, the strength used in testing. Reduce it if the style changes the character or clothing more than you want.
- For a fully clothed result, describe the clothing and coverage explicitly. The LoRA can change necklines and outfit details.
- Use the version matching your base model: Illustrious, Anima, Krea 2 or MOSS Image.
Version notes
- Illustrious: Clear painted cartoon styling. The 1,000-step checkpoint followed clothing details more closely in testing; the 2,000-step version introduced more fantasy features and wardrobe changes.
- Anima: Dimensional shading and expressive illustrated faces. The 2,000-step checkpoint gave the better overall result in the tested scenes.
- Krea 2: Strong style transfer, but generated logos, lettering and signatures appeared repeatedly during testing. Check the image carefully before using it.
- MOSS Image: Clear styling in portraits and individual character scenes. The effect was weaker in the tested café scene with two people.
Known limitations The dataset leans toward female characters and fantasy illustration. Faces, apparent age, clothing, poses and composition can shift along with the style. Hands and anatomy still need checking, and style strength varies with the scene and base model.