AI Diffusion Simulator

Image Diffusion
Simulator

Pick your favorite cute animal and watch the denoising diffusion model transform random noise into a stunning image

🐾 Choose a Cute Animal
Select an animal to begin
Output
Noise Level 100%
Diffusion Steps 50
Guidance Scale (CFG) 7.5
Progress 0 / 50
Denoising Steps Timeline
🔬

Forward Diffusion

Gradually adds Gaussian noise to an image until it becomes pure noise. This is the training phase of the model.

🎨

Reverse Diffusion

A neural network learns to reverse the noise process, gradually removing noise step by step to generate a new image.

Classifier-Free Guidance

CFG scale controls how strongly the model follows the prompt. Higher values produce more prompt-adherent but less diverse images.

🧠

U-Net Architecture

An encoder-decoder architecture with skip connections that predicts noise at each timestep for the denoising process.