How vladmandic/sdnext works
SD.Next is a cross-platform WebUI and server for generating, editing, and processing AI images and video. The documented user journey is to submit a generation or editing workflow through the interface/API, run model inference, then refine or process the result. The graph preserves README capabilities including captioning, ControlNet, LoRA, upscaling, model downloads, quantization, and installer support. Some wiring is based on README statements or SOURCE INDEX references; unsampled modules are represented at directory level, and uncertain runtime links are omitted.
Main entry points: processing.py Generation engine · sd_models.py Model loading · api Web UI and API · nr_pipeline.py DLSS features
Components 26
Interface and API
- Web UI and API
modules/api - Generation API
modules/api/generate.py - Generation history
modules/history.py
Generation and Editing
- Generation engine
modules/processing.py - Image-to-image
modules/img2img.py - APG pipelines
modules/apg/pipeline_stable_diffision_xl_apg.py - Control guidance
modules/control - LoRA adapters
modules/lora - Stable Diffusion APG
modules/apg/pipeline_stable_diffusion_apg.py
Media Processing
- Caption and tagging
modules/caption - Detailer
modules/detailer - Upscaling
modules/postprocess - Image processing
modules/image - Video workflows
modules/processing_video.py - DLSS features
modules/dlss/nr_pipeline.py - DLSS model
modules/dlss/nr_model.py - DLSS features
modules/dlss/nr_features.py
Models and Runtime
- Model loading
modules/sd_models.py - Model paths
modules/paths.py - SDNQ quantization
modules/model_quant.py - Attention backends
modules/attention - IPEX optimizations
modules/intel/ipex/diffusers.py
Platform Services
- Installer
installer.py - Device support
modules/devices.py - Logging
modules/logger.py
Other components
- Artist / API client
Connections 24
- Artist / API client → Web UI and API: submits workflow
README.md - Web UI and API → Generation API: offers generation
README.md - Generation API → Generation engine: runs generation
- Generation engine → Model loading: uses models
- Generation engine → Image-to-image: supports image editing
README.md - Generation engine → APG pipelines: supports diffusion pipeline
- Generation engine → Stable Diffusion APG: supports diffusion pipeline
- Generation engine → Control guidance: uses guidance
README.md - Generation engine → LoRA adapters: uses adapters
README.md - Generation engine → Caption and tagging: supports captioning
README.md - Generation engine → Video workflows: supports video
README.md - Generation engine → Detailer: supports refinement
README.md - Generation engine → Upscaling: supports upscaling
README.md - Generation engine → Image processing: supports processing
README.md - Web UI and API → Generation history: exposes history
- Model loading → Model paths: uses paths
- Model loading → SDNQ quantization: supports quantization
README.md - Generation engine → Attention backends: uses attention
- Generation engine → IPEX optimizations: uses optimizations
- Installer → Model loading: downloads models
README.md - Device support → Generation engine: enables execution
- DLSS features → DLSS model: uses model
modules/dlss/nr_pipeline.py - DLSS features → DLSS features: uses features
modules/dlss/nr_pipeline.py - Generation engine → DLSS features: supports enhancement