A LoRA model designed to transform any face into a digital caricature painting, with exaggerated proportions while retaining the subject’s identity.
Perfect for expressive, humorous, or satirical portraits that mix cartoon art and digital painting style.
Recommended prompt:
"C4ricature, a digital art painting caricature portrait of the [MAN/WOMAN/SUBJECT], retain and face features, exaggerated facial proportions while not changing his identity, retain pose, retain expression, [EXPRESSION] [Background/Scene]"
You also can do a full body caricature by replacing "Portrait" by "Full Body".
This LoRA enhances facial features (eyes, mouth, nose, brows) while trying to maintain recognizability. Results are often more effective with expressive faces or exaggerated emotions.
Close-up cropping on the face is highly recommended for best results.
You can change background, like : "gradiant background", "Shades of [color] background", etc...
V2 -> Trained over 5800 steps using AI-Toolkit on a 40-image custom dataset (Lr 0.0004)
Model is still evolving – V2 updates improve fidelity and structural consistency.
✦ Best suited for:
Creating humorous or satirical portraits.
Stylizing known faces while keeping their identity.
Making avatars, thumbnails, or eye-catching digital posters.
Description
FAQ
Comments (4)
Unfortunately, I was unable to install
Image Save, Latent Size to Number
This is the same thing than other Lora. Have you already use Flux Kontext ?
dodemodexter yes, but I couldn't find those two nodes I mentioned using your workflow
Z1nc0 don't use my workflow. It works with any "basic" workflow. You just have to add a Lora loader node between the checkpoint and the conditioner.