Nodes/ComfyUI-FunPack/FunPack Video Stitch
ComfyUI Node

FunPack Video Stitch

Stitch your scene clips into one timeline

By digital-garbage·Created about a year ago·Updated 2 days ago· 18
FunPack Video Stitch
  • video1
  • video2
  • video3
  • video4
  • video5
  • video6
  • video7
  • video8
  • STITCHED
blend_frames8
transition_typelinear

Every multi-scene video workflow ends the same way: you have several clips you generated separately and no graceful way to join them. FunPack Video Stitch is that join. It blends up to eight image-frame batches into one longer sequence, with proper transitions between neighboring clips - or a hard cut when you want one. It's the node you reach for when the scene chain has done its work and you need one continuous timeline to export.

How it works

Two settings govern the join. blend_frames (default 8) is how many frames each transition spans. Set it to 0 and the node cuts directly - no blending, clips concatenated end to end. transition_type picks the blend curve: linear, ease_in, ease_out, ease_in_out, or cosine. For most video work cosine is the smooth one and linear is the default you'll graduate from; the ease variants are for when you want the transition to land harder on one side.

Inputs video1..video8 are your clips in order. At least two connected inputs are required. Output is STITCHED - a single image batch you can hand to a video encoder (VideoHelperSuite is what the pack's own pipelines use) for the final file.

What to watch for

The one real constraint: each connected input must contain at least as many frames as blend_frames when transitions are enabled. A 4-frame clip with an 8-frame blend is a guaranteed error. If you're stitching clips of wildly different lengths, either keep blend_frames small or normalize lengths upstream.

The bigger conceptual point: this node blends in pixel space. It cross-fades the last N frames of one clip into the first N of the next. It does not make the underlying content match - if clip A ends with your character walking left and clip B starts with them facing the camera, you'll get a dissolve, not a cut that pretends they're the same take. Use it for cosmetic continuity (smoothing the seam between compatible clips), not for fixing incompatible ones.

Install

It's part of ComfyUI-FunPack:

cd ComfyUI/custom_nodes
git clone https://github.com/olivv-cs/ComfyUI-FunPack
pip install -r requirements.txt

or ComfyUI Manager → search "ComfyUI-FunPack". No extra dependencies - pure frame manipulation.

Where it fits

The natural partner is FunPack LTXAV Scene Chain Sampler, which generates the per-scene clips; Video Stitch then merges them into one timeline. In the Cutting Room editor the same idea is handled internally with real audio lanes, but in a classic graph this is your montage assembler. For the money, the combination of cosine transitions and a modest blend_frames is the look people are actually after - smooth, filmic seams instead of the visible jump cuts that betray separate generations.

CategoryFunPack

Inputs (10)

NameTypeDefaultDescription
blend_framesINT80–64
transition_typeCOMBOlinear5 options: linear, ease_in, ease_out, ease_in_out, cosine
video1optIMAGE
video2optIMAGE
video3optIMAGE
video4optIMAGE
video5optIMAGE
video6optIMAGE
video7optIMAGE
video8optIMAGE

Outputs (1)

NameTypeDescription
STITCHEDIMAGE