Qwen-Image-2512-Pixel-Art-LoRA

prithivMLmods/Qwen-Image-2512-Pixel-Art-LoRA

Qwen-Image-2512-Pixel-Art-LoRA is a text-to-image model from prithivMLmods. Use it when you need an image from a text prompt. It is set up for diffusers. The card lists the license as apache-2.0.

Downloads · 30 days

1.2K

19% of all-time downloads

All-time downloads

6.3K

Public

Repo size

1.2 GB

Likes

Trending 1

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors1.2 GB · 99%

At a glance

Task: Text-to-Image
Library: diffusers
License: apache-2.0
Access: Public
Created: Jan 1, 2026
Updated: Jan 5, 2026
SHA: e14eea58

Base models

Qwen/Qwen-Image-2512

Spaces

Model card

From the Hugging Face model README

pixel-art

Pixel Art, A pixelated image of a space astronaut floating in zero gravity. The astronaut is wearing a white spacesuit with orange stripes. Earth is visible in the background with blue oceans and white clouds, rendered in classic 8-bit style. Pixelated text reading “Qwen Pixel Art” appears at the bottom in a retro arcade font.

Qwen-Image-2512-Pixel-Art-LoRA

Image	Description
	Pixel Art, a pixelated image of a space astronaut floating in zero gravity. The astronaut wears a white spacesuit with orange stripes. Earth appears in the background with blue oceans and white clouds, rendered in classic 8-bit style.
	Pixel Art, a pixelated image of a film reel and projector. The projector emits a bright beam of light, casting moving frames onto a wall, styled like an old 8-bit video game.
	Pixel Art, a pixelated image of an ancient Greek temple with tall marble columns. Olive trees surround the temple, and the ground is made of stone tiles. Styled in detailed 16-bit pixel art.
	Pixel Art, a pixelated image of an ancient marketplace. Traders sell pottery, spices, and cloth. People in traditional robes walk through dusty paths between stalls, rendered in classic pixel art style.
	Pixel Art, a pixelated image of a man in a white tank top with dark brown hair and a black beard. He has a serious expression, with a blue sky and white clouds in the background.
	Pixel Art, a pixelated image of a man holding a cell phone. He wears a green sweater and blue jeans, with brown hair and brown eyes. The background is light beige.
	Pixel Art, a pixelated image of a black woman with a black afro. Her face is stylized using small white pixel lines. The background is light brown.
	Pixel Art, Retro Pixel style. A pixelated hamburger designed as a final boss in an old video game, with a hero-style composition and bold 8-bit colors.
	Pixel Art, a pixelated image of a German Shepherd dog with vibrant brown fur and a black stripe along its back. The background is light green, with a visible shadow on the ground.
	Pixel Art, a pixelated image of a man surfing on a vibrant blue surfboard. He wears a red shirt and blue shorts, with white water splashes and sunlight shining from the right.
	Pixel Art, Retro Pixel style. A pixelated farm at night with warm yellow light glowing from farmhouse windows. Stars fill the sky, and fireflies float above the fields.
	Pixel Art, a pixelated action movie hero walking away from an explosion. The hero wears a long coat and sunglasses, with flames and smoke rendered in dramatic 16-bit pixel style.

Model description for Qwen-Image-2512-Pixel-Art-LoRA

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	20 & 3000
Epoch	25	Save Every N Epochs	1

Labeling: prithivMLmods/DeepCaption-VLA-7B (natural language & English)

Total Images Used for Training : 50

Best Dimensions & Inference

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

Inference Range

Recommended Inference Steps: 45-50 / [Fused with Lightning/Turbo Adapters (4–8 Steps)]

Setting Up

import torch
from pipelines import DiffusionPipeline

base_model = "Qwen/Qwen-Image-2512"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)

lora_repo = "prithivMLmods/Qwen-Image-2512-Pixel-Art-LoRA"
trigger_word = "Pixel Art"  
pipe.load_lora_weights(lora_repo)

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

Trigger words

You should use Pixel Art to trigger the image generation.

Download model

Download them in the Files & versions tab.

Task: Text-to-Image
Library: diffusers
License: apache-2.0
Wavespeed: live
Created: Jan 1, 2026
Updated: Jan 5, 2026