Flux-Sketch-Sized-LoRA Open-Source Model - Free Deployment for Generating Sketch-Style Image Works

Flux Sketch Sized LoRA

Developed by strangerzonehf

A LoRA model fine-tuned based on the FLUX.1-dev base model, specializing in generating sketch-style artwork.

Image Generation Open Source License:Openrail #Sketch-style generation #Duotone rendering #Character portrait optimization

Downloads 85

Release Time : 12/21/2024

Model Overview

This model uses LoRA technology for lightweight fine-tuning of diffusion models, capable of generating sketch-style images based on text descriptions and supports various themes and styles.

Model Features

Sketch-style generation

Specially optimized for generating sketch-style artwork.

Duotone rendering

Excels at artistic rendering using 2-3 main tones.

Detail control

Capable of precisely controlling detail elements such as textures and expressions in images.

Model Capabilities

Text-to-image generation

Art style transfer

Detail-controlled generation

Use Cases

Artistic creation

Character portrait creation

Generate character portraits with specific styles and expressions.

Examples include character portraits in various styles.

Item concept design

Generate conceptual sketches of items.

Examples include sketches of robots, Christmas trees, and other items.

Entertainment applications

Fan art creation

Create artistic representations of popular culture characters.

Examples include sketches of popular culture elements like Iron Man helmets.

🚀 Flux-Sketch-Sized-LoRA

Flux-Sketch-Sized-LoRA is a text-to-image model based on LoRA and Diffusers, capable of generating various themed paintings according to text descriptions.

🚀 Quick Start

Environment Setup

import torch
from pipelines import DiffusionPipeline

base_model = "black-forest-labs/FLUX.1-dev"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)

lora_repo = "strangerzonehf/Flux-Sketch-Sized-LoRA"
trigger_word = "Sketch Sized"  
pipe.load_lora_weights(lora_repo)

device = torch.device("cuda")
pipe.to(device)

Image Generation

You should use Sketch Sized to trigger the image generation.

✨ Features

Rich Themed Images: Capable of generating various themed paintings, such as Wall-E, Christmas trees, Elon Musk's face, etc.
Customizable Parameters: Provides multiple image processing parameters for users to adjust according to their needs.

📦 Installation

The weights for this model are available in Safetensors format. Download them in the Files & versions tab.

💻 Usage Examples

Basic Usage

import torch
from pipelines import DiffusionPipeline

base_model = "black-forest-labs/FLUX.1-dev"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)

lora_repo = "strangerzonehf/Flux-Sketch-Sized-LoRA"
trigger_word = "Sketch Sized"  
pipe.load_lora_weights(lora_repo)

device = torch.device("cuda")
pipe.to(device)

# Generate an image
prompt = f"{trigger_word}, A medium-sized painting of Wall-E, painted in shades of yellow and black. His eyes, large and round, are painted in a glossy silver, giving a curious and endearing expression. His body is detailed with scratches and rust, hinting at years of service. The small, claw-like hands are slightly raised, as if reaching for something. The background is a soft sandy beige, with faint hints of scattered stars and dust."
image = pipe(prompt).images[0]
image.show()

📚 Documentation

Image Processing Parameters

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	27 & 3490
Epoch	20	Save Every N Epochs	1

Labeling

Labeling: florence2 - en(natural language & English)

Total Images Used for Training

Total Images Used for Training : 30 [ 14 bit raw ]

Best Dimensions & Inference

Dimensions	Aspect Ratio	Recommendation
1280 x 832	3:2	Best
1024 x 1024	1:1	Default

Inference Range

Recommended Inference Steps: 30–35

📄 License

This model is licensed under creativeml-openrail-m.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご