NEWSLETTER
In this tutorial, we will build a text generator application to an interactive image to which it was accessed through Google Colab and a public link using the Hugging Face and Graduate Diffusers Library. You will learn how to transform simple text indications into detailed images taking advantage of the latest generation stable diffusion model and GPU acceleration. We will pass the environment configuration, installing dependencies, storing the model in cache and creating an intuitive application interface that allows real -time parameter settings.
!pip install diffusers transformers accelerate gradioFirst, we install four essential python packages using PIP. Diffusers provide tools to work with diffusion models, Transformers offers prior models to tasks, accelerating optimizes performance in different hardware and graduation configurations allows the creation of interactive automatic learning interfaces. These libraries form the backbone of our demonstration of text generation in image on Google Colab. Set the execution time in GPU.
import torch
from diffusers import StableDiffusionPipeline
import gradio as gr
# Global variable to cache the pipeline
pipe = NoneNo, we import the necessary libraries: torch for tense calculations and acceleration of GPU, stabilization of Pipeline of the Diffusers Library to load and execute the stable and graduation diffusion model to build interactive demonstrations. In addition, a global variable pipe is initialized to none to store the pipe of the model loaded later, which helps avoid recharging the model in each inference call.
print("CUDA available:", torch.cuda.is_available())The previous code line indicates if there is a GPU enabled for CUDA available. Use the Torch.cuda.is_available () function of Pytorch, the function returns true if a GPU is detected and is ready for calculations and false, which helps to ensure that its code can take advantage of the acceleration of the GPU.
pipe = StableDiffusionPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5",
torch_dtype=torch.float16
)
pipe = pipe.to("cuda")The previous code fragment loads the stable dissemination pipe using a previously pretended model of “Runwayml/Stable-Diffusion-V1-5”. It establishes its data type at a floating point of 16 bits (torch.float16) to optimize the use and performance of memory. Then move the entire pipe to the GPU (“CUDA”) to take advantage of the acceleration of hardware for a faster generation of images.
def generate_sd_image(prompt, num_inference_steps=50, guidance_scale=7.5):
"""
Generate an image from a text prompt using Stable Diffusion.
Args:
prompt (str): Text prompt to guide image generation.
num_inference_steps (int): Number of denoising steps (more steps can improve quality).
guidance_scale (float): Controls how strongly the prompt is followed.
Returns:
PIL.Image: The generated image.
"""
global pipe
if pipe is None:
print("Loading Stable Diffusion model... (this may take a while)")
pipe = StableDiffusionPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5",
torch_dtype=torch.float16,
revision="fp16"
)
pipe = pipe.to("cuda")
# Use autocast for faster inference on GPU
with torch.autocast("cuda"):
image = pipe(prompt, num_inference_steps=num_inference_steps, guidance_scale=guidance_scale).images(0)
return image
The previous function, generate_sd_image, takes a text message together with the parameters for inference steps and guidance scale to generate an image using stable diffusion. Verify if the model pipe is already loaded in the global pipe variable; If not, load and store the model with half precision (FP16) and move it to the GPU. Then use torch. Autocast for efficient mixed precision inference and returns the generated image.
# Define the Gradio interface
demo = gr.Interface(
fn=generate_sd_image,
inputs=(
gr.Textbox(lines=2, placeholder="Enter your prompt here...", label="Text Prompt"),
gr.Slider(minimum=10, maximum=100, step=5, value=50, label="Inference Steps"),
gr.Slider(minimum=1, maximum=20, step=0.5, value=7.5, label="Guidance Scale")
),
outputs=gr.Image(type="pil", label="Generated Image"),
title="Stable Diffusion Text-to-Image Demo",
description="Enter a text prompt to generate an image using Stable Diffusion. Adjust the parameters to fine-tune the result."
)
# Launch the interactive demo
demo.launch()
Here, we define a graduate interface that connects the generate_sd_image function to an interactive web user interface. It provides three input widgets, a text box to enter the text message and sliding controls to adjust the number of inference steps and guidance scale. In contrast, the output widget shows the generated image. The interface also includes a title and descriptive text to guide users, and interactive demonstration finally begins.
You can also access the web application through a public URL: https://7dc6833297cf83b160.gradio.live/ (Active for 72 hours). A similar link will also be generated for its code.
In conclusion, this tutorial demonstrated how to integrate Hugging Face diffusers with graduates to create a powerful and interactive application of text in image in Google Colab and a web application. From the configuration of the GPU accelerated environment and storing in cache the stable diffusion model to build an interface for the dynamic interaction of the user, has a solid basis to experiment and develop even more advanced generative models.
Here is the Colab notebook For the previous project. Besides, don't forget to follow us <a target="_blank" href="https://x.com/intent/follow?screen_name=marktechpost” target=”_blank” rel=”noreferrer noopener”>twitter and join our Telegram channel and LINKEDIN GRsplash. Do not forget to join our 75K+ ml of submen.
<a target="_blank" href="https://www.marktechpost.com/2025/02/16/lg-ai-research-releases-nexus-an-advanced-system-integrating-agent-ai-system-and-data-compliance-standards-to-address-legal-concerns-in-ai-datasets/” target=”_blank” rel=”noreferrer noopener”>Recommended Reading Reading IA Research Liberations: An advanced system that integrates the ai system and data compliance standards to address legal concerns in IA data sets
Asif Razzaq is the CEO of Marktechpost Media Inc .. as a visionary entrepreneur and engineer, Asif undertakes to take advantage of the potential of artificial intelligence for the social good. Its most recent effort is the launch of an artificial intelligence media platform, Marktechpost, which stands out for its deep coverage of automatic learning and deep learning news that is technically solid and easily understandable by a broad audience. The platform has more than 2 million monthly views, illustrating its popularity among the public.
