CivArchive
    Kintsugi Art & Interior Concept | Z-IMAGE LoRA (Rank 128) - v1.0
    NSFW
    Preview 144280219
    Preview 144280352
    Preview 144280379
    Preview 144280416
    Preview 144280432
    Preview 144280461
    Preview 144280662
    Preview 144280683
    Preview 144280705
    Preview 144280731
    Preview 144281147
    Preview 144281146
    Preview 144281149

    A high capacity concept LoRA trained on unquantized, pure model weights (No Quantization / Full Precision FP8/bf16). This model captures the traditional Japanese art of Kintsugi (金継ぎ) — repairing broken ceramics with precious gold lacquer extending the aesthetics into wall art, fine porcelain, and elegant modern interior design.

    Key Features & Training Specs

    • Training Rank: Rank 128 (captures deep artistic gestures, complex gold textures, and exact light reflections).

    • Precision: 100% Unquantized base weights. No precision compression, no compromise on detail.

    • Focus: Traditional Kintsugi ceramics (vases, cups, plates) + Concept Interior Design (wall art paintings, gallery-style rooms, organic gold aesthetic).

    • Sampler: UniPC

    • Scheduler: Simple / SG Uniform

    • Steps: 50

    • CFG Scale: 4.0

    • LoRA Weight: 0.8 – 1.0

    Description

    LORA
    ZImageBase

    Details

    Downloads
    34
    Platform
    CivitAI
    Platform Status
    Available
    Created
    9/30/2026
    Updated
    10/10/2026
    Deleted
    -
    Trigger Words:
    r128zkitsu

    Files

    r128zkitsu.safetensors

    Mirrors

    HuggingFace (1 mirrors)
    CivitAI (1 mirrors)