---
pipeline_tag: text-to-image
widget:
- text: >-
   The image portrays a female character with vibrant blue hair, striking makeup, and tattoos. She is seen in a moment of distress or passion, with her eyes closed and one hand on her head. The background is a blend of purple hues with abstract symbols and doodles, giving an artistic and chaotic feel. The character is adorned with multiple accessories, including chokers, earrings, and a necklace. Her attire consists of a harness-like vest with various pouches and belts. The overall mood of the image is intense and evocative.
  output:
    url: examples/900M_OriNoiseSch_3.png
- text: >-
    Close-up photos of models, hazy light and shadow, laser metal hair accessories, soft and beautiful, light gold pupils, white eyelashes, low saturation, real skin details, clear pores and fine lines, light reflection and refraction, ultra-clear, cinematography, award-winning works
  output:
    url: examples/900M_OriNoiseSch_6.png
- text: >-
    A curvy timber house near a sea, designed by Zaha Hadid, represent the image of a cold, modern architecture, at night, white lighting, highly detailed
  output:
    url: examples/900M_OriNoiseSch_7.png
- text: >-
    Over the shoulder game perspective, game screen of Diablo 4, Inside the gorgeous palace is the wet ground, The necromancer knelt before the king, and a horde of skeletons he summoned stood at his side, cinematic light.
  output:
    url: examples/900M_OriNoiseSch_8.png
- text: >-
    Photorealistic closeup video of two pirate ships battling each other as they sail inside a cup of coffee.
  output:
    url: examples/900M_OriNoiseSch_12.png
- text: >-
    A car made out of vegetables.
  output:
    url: examples/900M_OriNoiseSch_13.png

- text: >-
    A deep forest clearing with a mirrored pond reflecting a galaxy-filled night sky.
  output:
    url: examples/900M_OriNoiseSch_14.png

- text: >-
    Happy dreamy owl monster sitting on a tree branch, colorful glittering particles, forest background, detailed feathers.
  output:
    url: examples/900M_OriNoiseSch_15.png

- text: >-
    Lego model, future rocket station, intricate details, high resolution, unreal engine, UHD
  output:
    url: examples/900M_OriNoiseSch_16.png
    
license: apache-2.0
---
<Gallery />

# PixArt-900M-DiT v0.1

This is an experimental 900M DiT model that has been fully fine-tuned on a large-scale, high-quality dataset of images aligned with human preferences. It uses the same VAE and text encoder as PixArt-Sigma, so you only need to replace the transformer.

# How to use it?
```python
weight_dtype = torch.bfloat16
device = torch.device("cuda")
MODEL_ID = "PixArt-alpha/pixart_sigma_sdxlvae_T5_diffusers"
transformer = Transformer2DModel.from_pretrained("Owen777/pixart-900m", torch_dtype=weight_dtype) 
pipe = PixArtSigmaPipeline.from_pretrained(MODEL_ID, transformer=transformer, torch_dtype=weight_dtype) 
pipe.vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=weight_dtype).cuda()
pipe.to(device)
image = pipe("a cute cat with pure white background.",guidance_scale=3.5,num_inference_steps=30,height=1024,width=102)[0]
image.save("result.jpeg")

```


# Credits
Original PixArt Sigma model by PIXART-Σ (https://pixart-alpha.github.io/PixArt-sigma-project/) 

This model is created by Tian Ye and Peijie Dong from ROAS/DSA, HKUST-GZ.