diff --git a/js/docs/Sage_AdvSamplerInfo.md b/js/docs/Sage_AdvSamplerInfo.md index 9ad9f22..35cfdca 100644 --- a/js/docs/Sage_AdvSamplerInfo.md +++ b/js/docs/Sage_AdvSamplerInfo.md @@ -2,11 +2,33 @@ **Adv Sampler Info** -Outputs advanced sampler settings for more granular control. +Provides advanced sampler settings for granular control over the sampling process. -## Parameters -- **sampler**: The sampler to use. -- **advanced_settings**: Advanced options for the sampler. +## Inputs + +- **add_noise** (BOOLEAN, required): Whether to add noise during sampling + - Default: True +- **start_at_step** (INT, required): Step number to start sampling from + - Default: 0, Range: 0 to 10,000 +- **end_at_step** (INT, required): Step number to end sampling at + - Default: 10,000, Range: 0 to 10,000 +- **return_with_leftover_noise** (BOOLEAN, required): Whether to return result with remaining noise + - Default: False + +## Outputs + +- **ADV_SAMPLER_INFO**: Dictionary containing advanced sampler parameters ## Usage -Use when you need to expose or control additional sampler parameters. + +Use when you need granular control over the sampling process beyond basic parameters. Connect to KSampler nodes to enable advanced sampling features like partial denoising, step ranges, and noise control. + +## Notes + +- Designed to work with Sage_KSampler and related sampler nodes +- `add_noise`: Set to False for img2img workflows where noise should not be added +- `start_at_step` and `end_at_step`: Enable partial sampling for multi-stage workflows +- `return_with_leftover_noise`: Useful for multi-pass sampling or when preserving noise structure +- Step range allows for sampling only specific portions of the denoising process +- Provides functionality similar to ComfyUI's advanced KSampler +- Optional input for sampler nodes - basic functionality works without it diff --git a/js/docs/Sage_KSampler.md b/js/docs/Sage_KSampler.md index eeef23c..6fa8703 100644 --- a/js/docs/Sage_KSampler.md +++ b/js/docs/Sage_KSampler.md @@ -2,11 +2,38 @@ **KSampler w/ Sampler Info** -A KSampler node with sampler settings broken out for easier workflow integration. Can be connected to Sampler Info nodes for streamlined configuration. +A KSampler node designed to work with Sampler Info nodes for streamlined configuration and workflow integration. -## Parameters -- **model**: The model to sample from. -- **sampler_info**: Sampler settings input. +## Inputs + +### Required + +- **model** (MODEL): The model used for denoising the input latent +- **sampler_info** (SAMPLER_INFO): Sampler settings from Sage_SamplerInfo node +- **positive** (CONDITIONING): Conditioning describing attributes to include in the image +- **negative** (CONDITIONING): Conditioning describing attributes to exclude from the image +- **latent_image** (LATENT): The latent image to denoise +- **denoise** (FLOAT): Amount of denoising applied + - Default: 1.0, Range: 0.0 to 1.0, Step: 0.01 + +### Optional + +- **advanced_info** (ADV_SAMPLER_INFO): Advanced sampler options from Sage_AdvSamplerInfo node + +## Outputs + +- **LATENT**: The denoised latent image ## Usage -Use for workflows that require flexible sampler configuration and metadata integration. + +Use for workflows that require flexible sampler configuration and metadata integration. Connect a Sage_SamplerInfo node to provide centralized sampler settings, and optionally connect Sage_AdvSamplerInfo for advanced control options. + +## Notes + +- Designed to work with Sage_SamplerInfo for centralized configuration +- Supports advanced sampling options when Sage_AdvSamplerInfo is connected +- Lower denoise values maintain structure for image-to-image workflows +- Sampler info should be routed to both this node and metadata construction +- Uses ComfyUI's common_ksampler internally for compatibility +- Advanced info enables step ranges, noise control, and leftover noise options +- Streamlines workflow by separating sampler configuration from execution diff --git a/js/docs/Sage_KSamplerAudioDecoder.md b/js/docs/Sage_KSamplerAudioDecoder.md index 2e4129b..3684ab8 100644 --- a/js/docs/Sage_KSamplerAudioDecoder.md +++ b/js/docs/Sage_KSamplerAudioDecoder.md @@ -2,11 +2,42 @@ **KSampler + Audio Decoder** -A KSampler node with an audio decoder variant. Outputs both latent and audio data. +A specialized KSampler node designed for audio generation workflows. It performs sampling on latent audio data and automatically decodes it to audio format, outputting both the denoised latent and decoded audio. -## Parameters -- **model**: The model to sample from. -- **sampler_info**: Sampler settings input. +## Inputs + +### Required +- **model** (MODEL): The model used for denoising the input latent +- **sampler_info** (SAMPLER_INFO): Most of the KSampler options. Should be piped both here and to the Construct Metadata node +- **positive** (CONDITIONING): The conditioning describing the attributes you want to include in the audio +- **negative** (CONDITIONING): The conditioning describing the attributes you want to exclude from the audio +- **latent_audio** (LATENT): The latent audio to denoise +- **vae** (VAE): The VAE used for decoding the latent audio +- **denoise** (FLOAT): The amount of denoising applied, lower values will maintain the structure of the initial audio allowing for audio to audio sampling (default: 1.0, range: 0.0-1.0) + +### Optional + +- **advanced_info** (ADV_SAMPLER_INFO): Optional. Adds in the options an advanced KSampler would have + +## Outputs + +- **LATENT**: The denoised latent +- **AUDIO**: The decoded audio (44.1kHz sample rate with normalized waveform) ## Usage -Use for workflows that require audio decoding from latent representations. + +This node is specifically designed for audio generation workflows using latent diffusion models. It combines the sampling and decoding steps into a single node for convenience. + +Key features: + +- Automatic audio normalization (scales by 5x standard deviation, minimum 1.0) +- Fixed 44.1kHz sample rate output +- Supports advanced sampling options when connected to Sage_AdvSamplerInfo +- Works with the Sage_SamplerInfo node for consistent workflow integration + +## Notes + +- The audio output includes automatic normalization to prevent clipping +- This node is optimized for audio generation workflows +- Use with audio-specific VAE models for best results +- The latent_audio input should come from audio-compatible latent sources diff --git a/js/docs/Sage_KSamplerTiledDecoder.md b/js/docs/Sage_KSamplerTiledDecoder.md index dbac278..afd8d4e 100644 --- a/js/docs/Sage_KSamplerTiledDecoder.md +++ b/js/docs/Sage_KSamplerTiledDecoder.md @@ -2,12 +2,43 @@ **KSampler + Tiled Decoder** -A KSampler node with a built-in tiled VAE decoder. Outputs both a latent and an image. If a Tiling Info node is connected, performs tiled VAE decoding. +A KSampler node with integrated VAE decoder that supports optional tiled decoding for memory-efficient processing of large images. -## Parameters -- **model**: The model to sample from. -- **sampler_info**: Sampler settings input. -- **tiling_info** (optional): Tiling information for tiled decoding. +## Inputs + +### Required + +- **model** (MODEL): The model used for denoising the input latent +- **sampler_info** (SAMPLER_INFO): Sampler settings from Sage_SamplerInfo node +- **positive** (CONDITIONING): Conditioning describing attributes to include in the image +- **negative** (CONDITIONING): Conditioning describing attributes to exclude from the image +- **latent_image** (LATENT): The latent image to denoise +- **vae** (VAE): The VAE used for decoding the latent image +- **denoise** (FLOAT): Amount of denoising applied + - Default: 1.0, Range: 0.0 to 1.0, Step: 0.01 + +### Optional + +- **tiling_info** (TILING_INFO): Tiling parameters from Sage_TilingInfo node for memory-efficient decoding +- **advanced_info** (ADV_SAMPLER_INFO): Advanced sampler options from Sage_AdvSamplerInfo node + +## Outputs + +- **LATENT**: The denoised latent image +- **IMAGE**: The decoded image (tiled if tiling_info provided) ## Usage -Use for workflows that require tiled decoding and flexible sampler configuration. + +Use for workflows that require both sampling and immediate image output with optional memory-efficient tiled decoding. Connect Sage_TilingInfo to enable tiled VAE decoding for large images, and optionally connect Sage_AdvSamplerInfo for advanced sampling control. + +## Notes + +- Combines sampling and VAE decoding in a single node +- Automatically uses tiled decoding when Sage_TilingInfo is connected +- Tiling enables processing of high-resolution images with limited VRAM +- Supports both spatial and temporal tiling for video content +- Advanced sampling options available through Sage_AdvSamplerInfo +- Automatically adjusts tiling parameters for compatibility +- Handles video batch reshaping for proper output format +- More efficient than separate sampler + decoder nodes +- Essential for high-resolution workflows with VRAM constraints diff --git a/js/docs/Sage_SamplerInfo.md b/js/docs/Sage_SamplerInfo.md index 7f57593..605e261 100644 --- a/js/docs/Sage_SamplerInfo.md +++ b/js/docs/Sage_SamplerInfo.md @@ -2,11 +2,37 @@ **Sampler Info** -Outputs sampler settings for use in metadata or workflow logic. +Collects and packages sampler settings for use in metadata construction and KSampler nodes. -## Parameters -- **sampler**: The sampler to use. -- **settings**: Additional sampler settings. +## Inputs + +- **seed** (INT, required): Random seed for noise generation + - Default: 0, Range: 0 to 2^64-1 +- **steps** (INT, required): Number of denoising steps + - Default: 20, Range: 1 to 10,000 +- **cfg** (FLOAT, required): Classifier-Free Guidance scale + - Default: 5.5, Range: 0.0 to 100.0, Step: 0.1 +- **sampler_name** (dropdown, required): Sampling algorithm + - Default: "dpmpp_2m" + - Options: All available ComfyUI samplers +- **scheduler** (dropdown, required): Noise scheduling algorithm + - Default: "beta" + - Options: All available ComfyUI schedulers + +## Outputs + +- **SAMPLER_INFO**: Dictionary containing all sampler parameters ## Usage -Connect to samplers or metadata nodes to standardize and share sampler configuration. + +Connect to both the Construct Metadata node and KSampler nodes to standardize and share sampler configuration across your workflow. This node centralizes sampler settings, making it easy to maintain consistency and modify parameters in one place. + +## Notes + +- Designed to work with Sage_KSampler and other sampler nodes +- Output should be routed to both metadata construction and sampling nodes +- CFG scale balances creativity and prompt adherence +- Higher CFG values increase prompt adherence but may reduce quality if too high +- Sampler algorithm affects quality, speed, and style of generation +- Scheduler controls how noise is gradually removed during denoising +- Centralizes sampler configuration for workflow consistency diff --git a/js/docs/Sage_TilingInfo.md b/js/docs/Sage_TilingInfo.md index f7407d8..346f422 100644 --- a/js/docs/Sage_TilingInfo.md +++ b/js/docs/Sage_TilingInfo.md @@ -2,11 +2,35 @@ **Tiling Info** -Outputs tiling information for tiled decoding or processing. +Provides tiling parameters for memory-efficient processing of large images and videos. -## Parameters -- **tile_size**: Size of each tile. -- **overlap**: Overlap between tiles. +## Inputs + +- **tile_size** (INT, required): Size of each spatial tile in pixels + - Default: 512, Range: 64 to 4096, Step: 32 +- **overlap** (INT, required): Overlap between spatial tiles in pixels + - Default: 64, Range: 0 to 4096, Step: 32 +- **temporal_size** (INT, required): Number of frames to process at once (video VAEs only) + - Default: 64, Range: 8 to 4096, Step: 4 +- **temporal_overlap** (INT, required): Number of frames to overlap (video VAEs only) + - Default: 8, Range: 4 to 4096, Step: 4 + +## Outputs + +- **TILING_INFO**: Dictionary containing all tiling parameters ## Usage -Connect to nodes that support tiled processing, such as tiled decoders. + +Connect to tiled decoder nodes like Sage_KSamplerTiledDecoder to enable memory-efficient processing of large images or videos. Tiling breaks down large content into smaller chunks that can be processed with limited VRAM. + +## Notes + +- Designed for use with Sage_KSamplerTiledDecoder and similar tiled processing nodes +- Spatial tiling (tile_size/overlap) applies to all content types +- Temporal tiling (temporal_size/temporal_overlap) only used for video VAEs +- Larger tile sizes use more VRAM but may produce better quality +- Overlap helps reduce seam artifacts between tiles +- System automatically adjusts parameters if they're incompatible (e.g., overlap too large) +- Temporal parameters are scaled based on VAE compression ratios +- Essential for processing high-resolution images with limited VRAM +- Optional input - nodes work without tiling if not provided