A LoRA that converts photos, 3D renders and realistic images into a western visual-novel comic style inspired by the artwork seen in Trials in Tainted Space.
Works especially well with:
photos
3D renders
realistic characters
semi-realistic anime models
The model focuses on clean lineart, strong color separation and simplified comic shading while avoiding the exaggerated anatomy typical for the original style.
Dataset generation process
Training images were created using a multi-stage pipeline:
Base images generated from photos and 3D renders
Stylized using a custom workflow based on the FLUX image generation model
Images were manually edited and composited to introduce backgrounds and scene context
Anatomy was adjusted to reduce exaggerated proportions and create a more neutral visual-novel style
The resulting dataset is a stylized derivative dataset, not a collection of original artworks.
This model is inspired by western visual novel artwork styles similar to those seen in Trials in Tainted Space, but the training dataset consists of synthetic and heavily modified images.