Thuật ngữ AI
Từ điển đầy đủ về Trí tuệ nhân tạo
Latent Diffusion Models
Diffusion architecture operating in a compressed latent space to reduce computational costs while preserving generative quality.
Text-to-Image Generation
Systems generating images from text descriptions using text-guided diffusion models.
Image-to-Image Translation
Translation of a source image into a target image while preserving semantic structure via conditional diffusion models.
Stable Diffusion
Open-source diffusion model based on latent space enabling high-resolution image generation with granular control.
3D Diffusion Models
Generation of three-dimensional content (meshes, point clouds, NeRFs) by applying diffusion principles to volumetric data.
Video Generation with Diffusion
Creation of temporally coherent video sequences by extending diffusion models to spatio-temporal data.
Audio Generation with Diffusion
Synthesis of sounds, music, and speech by applying diffusion processes to audio spectro-temporal representations.
ControlNet and Control Techniques
Methods enabling precise control over image generation through spatial conditions such as poses, edges, or depth maps.
Fine-tuning of Diffusion Models
Techniques for personalized adaptation of pre-trained models on specific data or particular artistic styles.
Conditional Diffusion Models
Generative systems controlled by multiple conditions (text, image, mask, class) for targeted generation.
Diffusion Models for Medical Imaging
Application of diffusion models to medical image synthesis and enhancement for diagnosis and research.
Optimization of Diffusion Models
Acceleration and computational efficiency techniques (distillation, pruning, quantization) for production deployment.