Parameter	Value	Parameter	Value
LR Scheduler	constant	Noise Offset	0.03
Optimizer	AdamW	Multires Noise Discount	0.1
Network Dim	64	Multires Noise Iterations	10
Network Alpha	32	Repeat & Steps	18 & 1600
Epoch	10	Save Every N Epochs	1

Labeling: florence2-en(natural language & English)

Total Images Used for Training : 20 [ Hi-RES ]

Sample Image + Prompt

Prompt

A bright green leaf logo is depicted against a dark blacked background. The leaf has smooth, organic lines with subtle veins running through it, giving it a natural and lifelike appearance. A soft glow surrounds the leaf, adding a gentle illumination that contrasts beautifully with the dark background. This contrast between the vibrant green and deep black creates a fresh and eco-friendly feel, making the logo appear both vibrant and serene.

Setting up

python

import torch
from diffusers import StableDiffusion3Pipeline

pipe = StableDiffusion3Pipeline.from_pretrained("stabilityai/stable-diffusion-3.5-large", torch_dtype=torch.bfloat16)
pipe.load_lora_weights("prithivMLmods/SD3.5-Large-Minimal-Blacked-LoRA", weight_name="SD3.5-Large-Minimal-Blacked-LoRA.safetensors")
pipe.fuse_lora(lora_scale=1.0)
pipe.to("cuda")

prompt = "A bold silver gear logo is depicted against a solid blacked background. The gear has sharp, metallic edges and intricate inner details that give it a realistic, industrial look. Light reflections on the gear add a gleaming effect, making it appear polished and three-dimensional. The stark contrast between the shiny silver and the dark black background creates a modern and powerful impression, giving the logo a sleek, mechanical feel."
negative_prompt = "(lowres, low quality, worst quality)"

image = pipe(prompt=prompt,
             negative_prompt=negative_prompt
             num_inference_steps=24, 
             guidance_scale=4.0,
             width=960, height=1280,
            ).images[0]
image.save(f"example.jpg")

Trigger words 🧨

[!WARNING] Trigger words: You should use blacked to trigger the image generation.

Download model

Weights for this model are available in Safetensors format.

Download them in the Files & versions tab.