FugitiveExpert01 — ComfyUI VFX Nodes A growing collection of custom ComfyUI nodes built for VFX production pipelines — designed around the real demands of working with high-resolution imagery, video sequences, and AI-assisted visual effects workflows.

Overview This repo aims to fill the gaps between off-the-shelf ComfyUI nodes and the specific needs of VFX work: large format images, temporal consistency across video frames, precise spatial control, and clean integration with diffusion-based upscaling and enhancement models. Nodes are built with video batch support as a first-class concern — not an afterthought.

Current Nodes TileSplit Splits an image or video batch into an overlapping grid of tiles, ready to be passed individually to a model (e.g. a video K-sampler or upscaler).

Configurable grid via tiles_x / tiles_y Overlap calculated automatically and stored for seamless reconstruction Outputs tiles as a list of (F, H, W, C) tensors — compatible with video model K-samplers Debug output image showing tile boundaries and overlap regions Tile dimensions snapped to multiples of 8 for model compatibility

Outputs: tiles (LIST), debug_image, tile_calc

TileMerge Reconstructs a full image or video sequence from processed tiles using Laplacian Pyramid blending for seamless, artifact-free joins at tile boundaries.

Gaussian feathering masks at overlap regions Laplacian frequency blending for natural-looking seams Robust handling of tensor shapes returned by video models Weighted accumulation handles overlapping regions correctly

Inputs: tiles (LIST), tile_calc

Installation

Navigate to your ComfyUI custom nodes directory:

ComfyUI/custom_nodes/

Clone this repository:

bashgit clone https://github.com/FugitiveExpert01/comfyui-vfx-nodes.git

Restart ComfyUI. Nodes will appear in the node menu under the VFX category.

No additional dependencies beyond what ComfyUI already requires (PyTorch, NumPy, Pillow).

Typical Workflow Load Video → TileSplit → [Video Model / K-Sampler per tile] → TileMerge → Save Video Because TileSplit outputs a list of tile sequences, you can route each tile through a video model independently and then reconstruct the full frame with TileMerge. This allows processing of resolutions that would otherwise exceed VRAM limits.

Roadmap This repo is actively developed alongside VFX production work. Planned and in-progress nodes include:

Optical Flow Warp — frame-to-frame warping for temporal consistency Depth-Guided Composite — layer compositing driven by estimated depth maps Multi-pass Denoise — frequency-separated denoising passes for fine/coarse detail Temporal Blend — blending across frame windows to reduce flicker Matte Operations — erode, dilate, blur, and combine alpha mattes Color Space Utilities — linear/log/sRGB conversions for VFX-accurate colour handling

Have a node idea or a production use case that isn't covered? Open an issue.

Contributing Pull requests are welcome. If you're adding a node, please:

Keep video batch support (F, H, W, C) as the primary tensor convention Add a docstring describing what the node does and its input/output types Test with both single images and multi-frame video batches

License MIT License — free to use in personal and commercial VFX pipelines.