Diffusion Model
Definition
A generative neural network architecture (Ho et al., 2020; Stable Diffusion, Rombach et al., 2022) that learns to reverse a noise-addition process applied to training images. Produces photorealistic image synthesis and inpainting without the mode-collapse instability of GANs. Adobe Firefly Generative Fill is a consumer-accessible implementation.
- Architecture type
- Generative neural network
- Key advantage over GANs
- Avoids mode-collapse instability, produces photorealistic results
- Capabilities
- Image synthesis and inpainting from text or image prompts
Common questions
How does a diffusion model generate an image?+
A diffusion model starts with random noise and progressively denoises it step by step, guided by a text prompt or existing image. Each step removes noise until a coherent, photorealistic image emerges that matches the prompt.
Why are diffusion models better than GANs for image generation?+
Diffusion models produce photorealistic images without the mode-collapse problem that affects GANs. Mode-collapse causes GANs to generate repetitive or limited outputs. Diffusion models learn to reverse the noise process directly, which is more stable.
What are real-world examples of diffusion models?+
Stable Diffusion and DALL-E 3 are well-known public implementations. Adobe Firefly Generative Fill is a consumer-accessible tool built on diffusion architecture that many people encounter for image editing and inpainting tasks.
Related terms
- Latent Space
- The compressed, lower-dimensional representation of data learned by a neural network's internal layers. Generative models sample from or navigate this space to...
- Adversarial Attack
- In AI detection contexts, a forgery specifically constructed to defeat a trained detector by incorporating small perturbations that cause the detector's classifier...
- Autoencoder
- A neural network that compresses an input into a compact latent representation (encoder) then reconstructs it (decoder). Face-swap pipelines train autoencoders with...
- Automatic Signature Verification (ASV)
- Algorithmic systems that compare a questioned signature against a reference set using extracted biometric features (stroke curvature, velocity profile, pressure distribution). Designed...
- Blending Mask
- In face-swap pipelines, a pixel-level mask that defines the face region to be composited onto the target frame. Imperfect masks leave boundary...
- C2PA (Coalition for Content Provenance and Authenticity)
- An open technical standard that embeds cryptographically signed provenance assertions into media files at the point of capture or editing. A C2PA...
- Checkerboard Artifact
- A grid-like pattern visible in the Fourier power spectrum of GAN outputs, caused by transposed convolution or bilinear-upsampling operations used to increase...
- Content-Aware Fill
- An image editing feature (Photoshop from CS5; AI-driven from version 22.0) that replaces a masked region with synthesised pixel content matching the...
- Encoder-Decoder
- A neural architecture where an encoder compresses an input into a compact latent representation and a decoder reconstructs an output image from...
- ESDA (Electrostatic Detection Apparatus)
- A non-destructive technique for detecting indented impressions on paper caused by writing pressure. A physical ink signature on paper produces detectable ESDA...
- GAN
- Generative Adversarial Network. A framework with two neural networks, a generator that creates synthetic data and a discriminator that tries to distinguish...
- GAN (Generative Adversarial Network)
- A neural architecture introduced by Goodfellow et al. (2014) consisting of a generator and discriminator trained in opposition. The generator learns to...
Explained in these topics
- AI-Generated Documents, Deepfake Signatures and Image ManipulationA generative neural network architecture (Ho et al., 2020; Stable Diffusion, Rombach et al., 2022) that learns to reverse a noise-addition process applied to t...
- Deepfake Generation: GANs, Diffusion, and Face-Swap PipelinesA generative model trained to reverse a step-by-step noise-addition process. Starting from random noise, a trained diffusion model progressively denoises towar...
- How Deepfakes Are Generated: GANs, Diffusion and Face Swap PipelinesA generative model trained to reverse a noise-addition process. During inference, the model starts from random Gaussian noise and iteratively removes noise ove...