entity-001/Error-lora-fire

PK_1 (Experimental / Failed Character LoRA)

Prompt

PK_1,1girl, solo, long_hair, breasts, looking_at_viewer, blush, smile, blue_eyes, large_breasts, white_background, thighhighs, hat, navel, jewelry, twintails, standing, full_body, swimsuit, pink_hair, earrings, bikini, detached_sleeves, pointy_ears, nail_polish, makeup, multicolored_clothes, argyle_clothes, jester_cap, argyle_legwear, neck_ruff, jester

Model description

Overview

This is an experimental LoRA trained as a first attempt to capture the character identity of PK_1.

Status: Failed / Unstable Experiment

• Outcome: The training run did not produce the expected facial and style consistency.
• Identified Issues: Style drift between 2D and 3D dataset sources, mixed anatomy tags, and overfitting during the initial steps.
• Purpose of this Repo: Kept publicly as an open research and development log to document progress, failure points, and iterative improvements before developing v2.

Usage

• Trigger Word: `PK_1`
• Recommended Base Model: Anima
• LoRA Weight: `0.6 - 0.8` (Experimental)

📢 Community Feedback & Help (Please save me! 😅)

Feel free to download, test, or experiment with this weight.

• If you find weird artifacts, bad anatomy, or know why it broke: Please leave a comment or report an issue!
• If you know a better way to clean the dataset or tune the hyperparameters: Any tips or pull requests are warmly welcomed!

(Honestly, I am still learning this whole training pipeline and kinda confused, so please send help if you know what you are doing! 🤣)

Trigger words

You should use PK_1 to trigger the image generation.

Download model

Download them in the Files & versions tab.