ComfyUI Node Runs on cloud
🅛🅣🅧 Tiled Fusion Sampler
A ComfyUI node in sampling with 16 inputs and 1 output.
🅛🅣🅧 Tiled Fusion Sampler
- model
- positive
- negative
- latents
- sigmas
- sampler
- vae
- output
◄seed42►
◄cfg1.0►
◄tile_width1024►
◄tile_height576►
◄overlap_frac0.50►
◄blend_var0.05►
◄grid_cycle1►
◄tile_frames0►
◄canvas_deviceauto►
Categorysampling
Inputs (16)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | The diffusion model to use. | |
| positive | CONDITIONING | Positive conditioning from the IC-LoRA guide node. Must carry keyframe_idxs / guide_attention_entries so each tile can be cropped to the LoRA's trained window. | |
| negative | CONDITIONING | Negative conditioning from the IC-LoRA guide node. | |
| latents | LATENT | The latent RETURNED BY the IC-LoRA guide node (it carries appended guide frames plus the noise mask). A bare empty video latent will not work. | |
| sigmas | SIGMAS | Denoise schedule. Must be descending and end at 0. Fusion walks this list one step at a time. | |
| sampler | SAMPLER | From KSamplerSelect (same type as SamplerCustomAdvanced). Discrete step samplers only: euler, euler_ancestral, heun, dpm_2, … History methods (DPM++ 2M, LMS) lose their multi-step memory because fusion owns the loop. Adaptive samplers cannot fuse. Ancestral methods add independent noise per tile and can seam. | |
| seed | INT | 420–18446744073709550000 | Seed for the single shared noise field across all tiles. |
| cfg | FLOAT | 1.00–20 | Classifier-free guidance scale. Distilled models typically use 1.0. |
| tile_width | INT | 1024256–4096 | Tile width in pixels, multiple of 32. Set this to the IC-LoRA's trained spatial window. Tiles smaller than the trained window make every tile see content at the wrong scale. |
| tile_height | INT | 576256–4096 | Tile height in pixels, multiple of 32. Match tile_width to the IC-LoRA's trained spatial window. |
| overlap_frac | FLOAT | 0.500.25–0.75 | Fraction of each tile overlapping its neighbor. Stay at 0.5 or above: below that a periodic grid appears on structured content. |
| blend_var | FLOAT | 0.050.01–0.2 | Variance of the Gaussian blend weights in the overlaps. Leave at 0.05 unless you are measuring seams. |
| grid_cycle | INT | 11–4 | Cycle shifted tile grids across steps. 1 = fixed grid (cheapest), 4 = thorough. Spatial only: cycling the temporal grid slides content in time. |
| tile_frames | INT | 00–1000 | Temporal window in pixel frames (97 for LTX). 0 = one extent. Other values are snapped to 8n+1. >0 needs LTXAddVideoICLoRAGuide with use_streaming on and the SAME tile_frames: that re-encodes a fresh guide per window and writes the plan this sampler reads. A whole-clip guide cannot be sliced, and windowing is disabled if the plan is missing. |
| vaeopt | VAE | Used to read the model's temporal and spatial downscale factors. If omitted, they are taken from the diffusion model (LTX defaults: 8 frames and 32 pixels per latent cell). | |
| canvas_deviceopt | COMBO | auto | Where to keep the full-canvas tensors and cached tile latents/noise/masks. auto = GPU if there is enough free memory, otherwise CPU. Processed conds stay on the GPU. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| output | LATENT | — |