CivArchive
    Digital painting caricature – KonText LoRA - v1.0
    NSFW
    Preview 89952152
    Preview 89952172
    Preview 89952180
    Preview 89952182
    Preview 89952185

    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

    LORA
    Flux.1 Kontext

    Details

    Downloads
    204
    Platform
    CivitAI
    Platform Status
    Available
    Created
    7/23/2025
    Updated
    4/27/2026
    Deleted
    -
    Trigger Words:
    make 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]

    Files

    caricature.safetensors

    Mirrors

    TensorFiles (1 mirrors)