Unverified50% confidenceFactExact time
Stable Diffusion 基于去噪扩散概率模型（DDPM）理论，将扩散过程从像素空间转移到低维潜在空间，由VAE负责编解码，CLIP模型将文本提示编码为语义向量
1
Sources
50%
Confidence
Long-term
Relevance
7/17/2026
First Seen
Sources
Related Claims
VerifiedStable Diffusion is based on the Latent Diffusion Model and generates high-quality images by performing a denoising process in a compressed latent space79% similarUnverifiedIn 2020, DDPM (Denoising Diffusion Probabilistic Models) pioneered a paradigm in the image generation field of achieving high-quality generation through progressive noise addition followed by progressive denoising.79% similarUnverifiedThe Latent Diffusion Model (LDM) is the architecture used by Stable Diffusion, in which original images are encoded by a VAE into latent vectors approximately 1/8 the size of the original image.76% similarUnverifiedThe theoretical foundation of diffusion models was unified by Song Yang et al. in 2020 through a stochastic differential equation (SDE) framework; the discovery of probability flow ODEs enabled deterministic sampling and gave rise to accelerated sampling methods such as DDIM, compressing the number of sampling steps from 1000 to 10–50.73% similarVerifiedStable Diffusion is built on the Latent Diffusion Model (LDM) architecture73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/544836API
curl https://kongchang.com/api/v1/knowledge/claims/544836MCP
get_claim(id=544836)