ComfyUI Node
Bernini-R Context Window
Temporal context windows for long video generation. Configures how the sampler splits video frames into overlapping windows for memory-efficient processing. Mirrors WanVideoWrapper's WanVideoContextOptions.
Bernini-R Context Window
- context_options
◄context_schedulestatic_standard►
◄context_frames81►
◄context_stride4►
◄context_overlap16►
◄freenoisetrue►
◄fuse_methodsmooth►
◄rope_ntk_scale1.00►
CategoryBernini-R/Sampling
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| context_schedule | COMBO | static_standard | Window scheduling: 'static_standard' = simple sliding (fastest), 'uniform_standard' = multi-stride (best quality), 'uniform_looped' = cyclic for seamless looping |
| context_frames | INT | 812–1000 | Pixel frames per window. 81 ≈ 20 latent frames. Smaller = less VRAM, more windows |
| context_stride | INT | 41–100 | Stride levels in log2 scale. Higher = more window positions (uniform_* only) |
| context_overlap | INT | 160–100 | Overlap between windows in pixel frames. Smooths transitions. 12-16 typical |
| freenoise | BOOLEAN | true | Shuffle noise across windows to reduce seam artifacts |
| fuse_methodopt | COMBO | smooth | Blending: 'smooth' (default) = smoothstep crossfade (softest seams), 'linear' = crossfade at edges, 'pyramid' = triangle weights peaking at center |
| rope_ntk_scaleopt | FLOAT | 1.001–8 | NTK RoPE frequency scaling for sequences longer than training. 1.0 = off. 2.0 = moderate (2x training length), 3.0+ = aggressive. Scales rope_embedder.theta to prevent positional collapse. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| context_options | BERNINI_CTX | — |