Man pages for sd2R
Stable Diffusion Image Generation
| dot-api_error | Build JSON error response |
| dot-array_to_sd_image | Convert R array [H, W, 3] to sd_image list |
| dot-base64_to_image | Decode base64 PNG to sd_image |
| dot-blend_mask | Build linear blend mask for a patch |
| dot-build_router | Build plumber router with sd2R endpoints |
| dot-compute_patch_grid | Compute patch grid positions |
| dot-estimate_vae_vram | Estimate peak VAE VRAM usage in bytes |
| dot-find_lru | Find least recently used model id |
| dot-get_ctx | Get model context by name (or default) |
| dot-guess_component | Guess component role from filename |
| dot-guess_model_type | Guess model type from filename |
| dot-images_to_base64 | Encode sd_image list to base64 PNG strings |
| dot-native_latent_tile_size | Get native latent tile size for a model type |
| dot-native_tile_size | Get native tile size for a model type |
| dot-resize_sd_image | Bilinear resize of an SD image |
| dot-resolve_device_layout | Resolve device layout preset to concrete GPU indices |
| dot-resolve_vae_tiling | Resolve VAE tiling mode to boolean |
| dot-select_strategy | Select generation strategy based on resolution and VRAM |
| dot-unbox_scalars | Recursively unbox scalar values in nested lists for JSON... |
| LORA_APPLY_MODE | LoRA apply modes |
| PREDICTION | Prediction types |
| RNG_TYPE | RNG types |
| SAMPLE_METHOD | Sampling methods |
| SCHEDULER | Schedulers |
| sd_api_start | Start sd2R REST API server |
| sd_api_stop | Stop sd2R REST API server |
| SD_CACHE_MODE | Cache modes |
| sd_cache_params | Create cache configuration for step caching |
| sd_convert | Convert model to different quantization format |
| sd_ctx | Create a Stable Diffusion context |
| sd_generate | Generate images (unified entry point) |
| sd_generate_multi_gpu | Parallel generation across multiple GPUs |
| sd_highres_fix | High-resolution image generation (Highres Fix) |
| sd_image_to_array | Convert SD image to R numeric array |
| sd_img2img | Generate images with img2img |
| sd_img2img_tiled | Tiled img2img (MultiDiffusion with init image) |
| sd_list_models | List registered models |
| sd_load_image | Load image from file as SD image |
| sd_load_model | Load a registered model |
| sd_load_pipeline | Load pipeline from JSON |
| sd_node | Create a pipeline node |
| sd_pipeline | Create a pipeline from nodes |
| sd_profile_get | Get raw profile events |
| sd_profile_start | Start profiling |
| sd_profile_stop | Stop profiling |
| sd_profile_summary | Build a profile summary from raw events |
| sd_register_model | Register a model in the sd2R model registry |
| sd_remove_model | Remove a model from the registry |
| sd_run_pipeline | Run a pipeline |
| sd_save_image | Save SD image to PNG file |
| sd_save_pipeline | Save pipeline to JSON |
| sd_scan_models | Scan a directory for models and register them |
| sd_system_info | Get system information |
| sd_txt2img | Generate images from text prompt |
| sd_txt2img_highres | High-resolution image generation via patch-based pipeline |
| sd_txt2img_tiled | Tiled diffusion sampling (MultiDiffusion) |
| SD_TYPE | Weight types (ggml quantization types) |
| sd_unload_all | Unload all models from memory |
| sd_unload_model | Unload a model from memory |
| sd_upscale_image | Upscale an image using ESRGAN |
| sd_vulkan_device_count | Get number of Vulkan GPU devices |
