Candy Sketchpop is a stylized character-focused LoRA for Anima, built to create cute and poppy anime characters with exaggerated facial expressions and a playful, eye-catching look.
It performs best on portraits and upper-body illustrations featuring a single character. This is where the style appears most consistent and visually appealing. The model is currently less suited for NSFW generations, scenes with multiple characters, or complex full-body posing.
The model was trained on 300 high-resolution images. It works well with both landscape and portrait-oriented outputs, as long as the shorter side remains in the 1024–1280 pixel range.
Recommended settings:
- Sampling method: Res Multistep - Simple
- Steps: 25–32
- CFG: 4
I also find that avoiding a negative prompt usually helps preserve the style more consistently.
Description
Initial release of Candy Sketchpop.
This first version was trained on 300 high-resolution images. It currently performs best on portraits and upper-body shots, where the style reads most cleanly and consistently.
The model is less reliable with complex scenes, multiple characters, and full-body poses.