ContextWindowsManual - ComfyUI Built-in Node Documentation

The Context Windows (Manual) node allows you to manually configure context windows for a model during sampling, creating overlapping context segments with a specified length, overlap, and scheduling pattern so data is processed in manageable chunks while maintaining continuity between segments. It provides advanced options for controlling how context windows are applied, including noise shuffling, conditioning retention, and causal window fixes. This node is experimental.

Inputs

Parameter	Description	Data Type	Required	Range
`model`	The model to apply context windows to during sampling.	MODEL	Yes	-
`context_length`	The length of the context window (default: 16).	INT	Yes	1+
`context_overlap`	The overlap of the context window (default: 4).	INT	Yes	0+
`context_schedule`	Step-dependent scheduling algorithm for context windows (default: STATIC_STANDARD).	COMBO	Yes	`STATIC_STANDARD` `UNIFORM_STANDARD` `UNIFORM_LOOPED` `BATCHED`
`context_stride`	The stride of the context window; only applicable to uniform schedules (default: 1).	INT	Yes	1+
`closed_loop`	Whether to close the context window loop; only applicable to looped schedules (default: False).	BOOLEAN	Yes	-
`fuse_method`	The method to use to fuse the context windows (default: PYRAMID).	COMBO	Yes	Static fuse methods (see `ContextFuseMethods.LIST_STATIC`)
`dim`	The dimension to apply the context windows to (default: 0).	INT	Yes	0-5
`freenoise`	Whether to apply FreeNoise noise shuffling, improves window blending (default: False).	BOOLEAN	Yes	-
`cond_retain_index_list`	List of latent indices to retain in the conditioning tensors for each window. For concat-style I2V models (e.g. Wan I2V, HunyuanVideo I2V, Cosmos I2V, SVD) the encoded start image lives in the c_concat conditioning channels; setting this to ‘0’ will retain that start image content at sub-pos 0 of every window (default: "").	STRING	No	-
`split_conds_to_windows`	Whether to split multiple conditionings (created by ConditionCombine) to each window based on region index (default: False).	BOOLEAN	No	-
`latent_retain_index_list`	List of latent indices to retain in the noise latent itself for each window. Use for workflows where reference content (e.g. a start image) lives directly in the noise latent rather than in separate conditioning channels (e.g. inplace-style I2V like LTXV, AnimateDiff). Independent of cond_retain_index_list (default: "").	STRING	No	-
`causal_window_fix`	Whether to add a causal fix frame to non-0-indexed context windows (default: True).	BOOLEAN	No	-

Parameter Constraints:

context_stride is only used when a uniform schedule is selected (UNIFORM_STANDARD or UNIFORM_LOOPED).
closed_loop is only applicable to looped schedules (UNIFORM_LOOPED).
dim must be between 0 and 5 inclusive.
cond_retain_index_list and latent_retain_index_list expect a comma-separated list of integer indices as a string (e.g., “0,1,2”).
latent_retain_index_list is independent of cond_retain_index_list.

Outputs

Output Name	Description	Data Type
`model`	The model with context windows applied during sampling.	MODEL

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 39dc39ece3d3c10c13ca8c4b85af4fbbebbcaba8a019145a6d4727c3df7b302b

​Inputs

​Outputs

Inputs

Outputs