---
license: creativeml-openrail-m
base_model: segmind/SSD-1B
dataset: recoilme/aesthetic_photos_xs
tags:
- stable-diffusion-xl
- stable-diffusion-xl-diffusers
- text-to-image
- diffusers
inference: true
---
    
# Text-to-image finetuning - recoilme/ColorfulSSD-1B_v02

This pipeline was finetuned from **segmind/SSD-1B** on the **recoilme/aesthetic_photos_xs** dataset. Below are some example images generated with the finetuned pipeline using the following prompt: girl, best quality, ultra detailed: 

![image/png](https://cdn-uploads.huggingface.co/production/uploads/63a2d82b528eba15b1392902/-dU74gW5j0M4BMGpQpFRJ.png)


Reproduce

```
from diffusers import StableDiffusionXLPipeline,UNet2DConditionModel,EulerAncestralDiscreteScheduler
import torch
from IPython.display import Image

pipe = StableDiffusionXLPipeline.from_pretrained("recoilme/ColorfulSSD-1B_v02", torch_dtype=torch.float16, use_safetensors=True)
pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
pipe.to("cuda")


neg_prompt = "deformed,low quality, worst quality, bad_quality, normal quality, cropped, fingers, distorted, disfigured, limb, hands, anatomy, long neck, blurry" # Negative prompt here
generator = torch.Generator("cuda").manual_seed(42)
steps=50

prompts = ["An astronaut riding a green horse", 
           "white snow covered mountain under blue sky during daytime", 
           "aerial view of the beach at night", 
           
           "a graphite sketch of Elon Musk",
          "painting of an alien by Claude Monet",
          "a background consisting of colors blue, white, and black",
          
          "best_quality, ultra_detailed, underwater, giant whale, fantastic location, dream, flying, underwater cyberpunk city",
          "best_quality, ultra_detailed, 8k, extremely_clear, photograph, running ninja, textured clothing, black background",
          "best_quality, ultra_detailed, 8k, portrait of beautiful cyborg with brown hair, intricate, elegant, highly detailed, majestic, digital photography, art by artgerm and ruan jia and greg rutkowski surreal painting gold butterfly filigree, broken glass",
          
          "photograph, beautiful, 1girl, scared, wide-eyed, shock, barely clothed, torn clothes, dress, multiple boys, zombie, motion blur, looking back, looking over his shoulder",
          "A cute pop singer with short hair and symmetrical teary eyes holds a guitar and sings into headphones, wearing a pleated skirt, with a dynamic angle, captured in a high-detailed cowboy shot, exuding a melancholy emotion, featuring soothing tones and a contrasting mix of light and shadow",
          "portrait of Asuka Evangelion, futuristic red reflective wingman latex suit, porclain skin, natural realistic Ginger hair, messy hair, pigtail, focus on eyes and face ,Rembrandt lighting, masterpiece:, best_quality, ultra_detailed, 8k, extremely_clear",
          
          "best_quality, ultra_detailed, A stunning woman with messy hair, flirty, sits in a classroom, artwork on the blackboard, short black skirt, unbuttoned white blouse, messy bun, camera above",
          "A half body, portrait of an Australian 21-year-old woman, captured in a solo shot, featuring a braid, blue eyes, small breasts, a naval, cropped legs, shackles, and a chain, with a leash restraining her, in a fantasy-themed scene",
          "best ratio, a photo of attractive stunning Ukrainian woman, messy bun, covered by snow, skin pores, behind glacial mountains, snow, high detailed skin, film grain, Fujifilm XT3, high detailed face, soothing tones, hdr, puffy nipples, topless, nsfw, flirty",
          
           "Fantastic location, bee and woman mix, photograph, highly detailed, sharp focus, 8k, 4k, hyperrealism, micro details, colorful, Beautiful environment, Portfolio piece, beautiful artwork",
          "black and white,solo, Pencil Sketch Drawing, 1girl, solo, little girl, black and white drawing, graphite drawing",
           "best_quality, ultra_detailed, bloom, road in forest, a close up of a wild forest black flowers, forest fantasy, anime nature, beautiful photo"
          
          ]
for i, prompt in enumerate(prompts):
    step = str(i + 1).zfill(3)
    image = pipe(prompt=prompt, negative_prompt=neg_prompt, generator=generator,num_inference_steps=steps).images[0]
    image.save(f"{step}.png")
```


Special VAE used for training: madebyollin/sdxl-vae-fp16-fix.