A LoRA model designed to transform your images into an expressive digital cartoon style, sitting at the crossroads between stylized illustration and 2.5D animation.
Use the prompt: "Turn this image to D1gitalCart00n style" to activate the effect.
This LoRA doesn’t just apply a filter — it reinterprets the scene. Characters are exaggerated, sometimes caricatured, with bold, intentional stylistic choices. The visual inspiration draws from animated series such as Gigantosaurus and Arcane.
The model was trained over 2500 steps on a local machine using AI-Toolkit, with a base dataset of 30 images.
✦ Best suited for:
Giving a narrative or stylized look to realistic scenes.
Turning iconic movie or TV visuals into illustrated versions.
Building training datasets for games or digital comics.
⚠️ Note:
The model often alters facial features and expressions significantly. It's not ideal for preserving realism or facial identity.
Description
FAQ
Comments (11)
Interesting, thanks!
What was your learning rate?
Cowboy_Gibop 0.0004
@dodemodexter
How long did it take you to train this Lora? Only 30 images. That's amazing.😎
Some few hours (ai toolkit but with 4 bits kontext model)
dodemodexter Thanks, also I am learning Flux IPAdapter to copy ref image style to another image :)
Very good Lora!!!!
Can you give me some advice on training lora,the learning rate,the trick of prompt?
Hi. Sure ! The LR was 0.0004. The very simple prompt was "Turn this image to D1gitalCart00n style" on each images.
dodemodexter will digitalcartoon style affect the effect of Lora? i see you replace it as D1gitalCart00n style
tqy965346240483 I think, but i'm not sure. To be honnest, i have not try. I usually put some 4 and 0 instead of letter in my triggers to be sure the style will be "pure". For exemplae, Post Apocalyptic in my other model is P0stApp0calyptic because the originals words "Post Apocalyptic" change the luminosity of the final result (too dark to me)
dodemodexter ok,thanks a lot! nice work!!!!!
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.