DDDDD-433/glowsticker-tiny-sd-lora

GlowSticker Tiny SD LoRA

Tiny LoRA adapter for segmind/tiny-sd trained on a synthetic PIL-generated sticker-art dataset.

Use the trigger phrase glowsticker style for best results.

Intended Use

This is a small public demo, not a production image model. It was built to show a complete tiny image-style LoRA workflow: synthetic data, HF Jobs training, adapter publishing, and a Gradio Space.

Training Data

• 72 synthetic images generated in the training job
• No real product data or scraped artwork
• Simple geometric objects: robot, mug, plant, moon, rocket, desk, and laptop
• Captions include bold sticker outline, soft cyan glow, flat pastel colors, dark grid background, and tiny sparkles

Example Prompts

• glowsticker style illustration of a tiny robot holding a coffee mug
• glowsticker style illustration of a standing desk with a little plant
• glowsticker style illustration of a moonlit coffee cup, tiny sparkles

Diffusers Usage

import torch
from diffusers import StableDiffusionPipeline

pipe = StableDiffusionPipeline.from_pretrained(
    "segmind/tiny-sd",
    torch_dtype=torch.float16,
    safety_checker=None,
    requires_safety_checker=False,
).to("cuda")
pipe.load_lora_weights("DDDDD-433/glowsticker-tiny-sd-lora")
image = pipe(
    "glowsticker style illustration of a friendly desk robot, soft cyan glow",
    num_inference_steps=12,
    guidance_scale=6.0,
    height=256,
    width=256,
).images[0]