Qwen-Image-Fragmented-Portraiture

Prompt

Fragmented Portraiture, a close-up shot of a young Asian girls face is seen through a transparent window. The girls head is tilted slightly to the left, and his eyes are wide open. Her hair is a vibrant shade of black, and he is wearing a white collared shirt with a white collar. Her lips are painted a bright pink, adding a pop of color to the scene. The backdrop is a stark white, creating a stark contrast to the boys body. The window is made up of thin, light-colored wooden blinds, adding depth to the image.

Prompt

Fragmented Portraiture, Captured in a black and white collage, a womans face is featured prominently in the center of the collage. The womans eyes are wide open, and her lips are pursed. Her hair is long and cascades over her shoulders. The background is a stark white, and the womans hair is a vibrant shade of brown, adding a pop of color to the composition.

Prompt

Fragmented Portraiture, Captured in a black and white monochrome, a close-up shot of a womans face is visible through a series of white vertical blinds. The womans eyes are wide open, and her lips are pursed. Her hair is long and cascades down to her shoulders, framing her face. The blinds are pulled up, adding a touch of depth to the scene. The background is a stark white, creating a stark contrast to the womans features.

---

Model description for Qwen-Image-Fragmented-Portraiture

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 & 3050
Epoch	20	Save Every N Epochs	2

Labeling: florence2-en(natural language & English)

Total Images Used for Training : 17 [HQ Images]

Data Sources

Source	Link
Playground	playground.com
ArtStation	artstation.com
4K Wallpapers	4kwallpapers.com

Best Dimensions & Inference

Dimensions	Aspect Ratio	Recommendation
1472 x 1140	4:3 (approx.)	Best
1024 x 1024	1:1	Default

Inference Range

• Recommended Inference Steps: 35-50

Setting Up

import torch
from diffusers import DiffusionPipeline

base_model = "Qwen/Qwen-Image"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)

lora_repo = "prithivMLmods/Qwen-Image-Fragmented-Portraiture"
trigger_word = "Fragmented Portraiture"  
pipe.load_lora_weights(lora_repo)

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

Trigger words

You should use Fragmented Portraiture to trigger the image generation.

Download model

Download them in the Files & versions tab.