How to beginner · 3 min read

How to use StableDiffusionPipeline

Q: How to use StableDiffusionPipeline

Use the StableDiffusionPipeline from the diffusers library to generate images from text prompts in Python. Install the library, load a pretrained model, and call the pipeline with your prompt to get generated images.

Quick answer

Use the StableDiffusionPipeline from the diffusers library to generate images from text prompts in Python. Install the library, load a pretrained model, and call the pipeline with your prompt to get generated images.

PREREQUISITES

Python 3.8+
pip install diffusers>=0.20.0
pip install torch (with CUDA for GPU acceleration recommended)
Hugging Face account and access token (if using private models)

Setup

Install the diffusers library and torch for model inference. Optionally, set your Hugging Face token as an environment variable for access to pretrained models.

bash

pip install diffusers torch

Step by step

This example shows how to load the Stable Diffusion model and generate an image from a text prompt.

python

from diffusers import StableDiffusionPipeline
import torch
import os

# Set your Hugging Face token if needed
# os.environ["HUGGINGFACE_TOKEN"] = "your_token_here"

# Load the pipeline with the pretrained model
pipe = StableDiffusionPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5",
    torch_dtype=torch.float16
)
pipe = pipe.to("cuda")  # Use GPU if available

# Generate an image from a prompt
prompt = "A beautiful sunset over mountains"
image = pipe(prompt).images[0]

# Save the image
image.save("output.png")
print("Image saved as output.png")

output

Image saved as output.png

Common variations

Use CPU by setting pipe.to("cpu") if no GPU is available.
Change model by specifying a different pretrained checkpoint in from_pretrained().
Generate multiple images by passing num_inference_steps and num_images_per_prompt parameters.
Use pipe(prompt, guidance_scale=7.5) to control image creativity and adherence to prompt.

python

images = pipe(prompt, num_inference_steps=50, num_images_per_prompt=3).images
for i, img in enumerate(images):
    img.save(f"output_{i}.png")
print("Saved 3 images")

output

Saved 3 images

Troubleshooting

If you get CUDA out of memory errors, reduce batch size or switch to CPU.
Ensure torch version matches your CUDA version.
If model download fails, verify your Hugging Face token and internet connection.
For slow generation, use GPU acceleration or optimize with torch_dtype=torch.float16.

✅

Key Takeaways

Use StableDiffusionPipeline from diffusers for easy text-to-image generation.
GPU acceleration with torch and float16 improves speed and reduces memory use.
Customize generation with parameters like num_inference_steps and guidance_scale.
Set up Hugging Face authentication for private or large models.
Troubleshoot memory and download issues by adjusting environment and dependencies.

Verified 2026-04 · runwayml/stable-diffusion-v1-5

Verify ↗