Skip to main content
Video Composite

Video Composite · screenshot coming soon

​
What this node does

Video Composite takes a background and a foreground video and merges them into a single clip. The foreground can be blended with any of the professional merge operators (over, screen, multiply, Porter-Duff ops, HSL modes, and more), faded with an opacity value, and repositioned/scaled/rotated in 2D before it lands on the background. An optional mask limits where the foreground shows, and an optional keyframe track animates the transform over time. It runs on ▶ Run (GPU/torch compositing). Both background and foreground are COMFYTV_VIDEO and must be wired — the node errors if either is missing. Output is a single COMFYTV_VIDEO. ComfyTV media flows as project snapshots, not native ComfyUI tensors. To hand the result to a stock ComfyUI node, insert a Bridge (ComfyTV/Bridge). See https://github.com/jtydhr88/ComfyTV/blob/main/docs/bridges.md.

​
When to use it

  • Drop a green-screen subject (already keyed) or a logo over a background plate.
  • Add a screen/plus glow layer, a multiply shadow, or a texture pass over footage.
  • Place a picture-in-picture insert and animate it flying in with keyframes.

​
Parameters

​
operator

The blend/merge mode used to combine foreground over background. Options (from MERGE_OPERATORS): atop, average, color, color-burn, color-dodge, conjoint-over, copy, difference, disjoint-over, divide, exclusion, freeze, from, geometric, grain-extract, grain-merge, hard-light, hue, hypot, in, luminosity, mask, matte, max, min, minus, multiply, out, over, overlay, pinlight, plus, reflect, saturation, screen, soft-light, stencil, under, xor. Default over (standard alpha composite).

​
opacity

Foreground opacity, 0.0–1.0, default 1.0. Fades the whole foreground layer toward transparent.

​
pos_x / pos_y

Translation of the foreground in pixels. Range -8192–8192, step 1, default 0. Positive pos_x moves right; positive pos_y moves down.

​
scale

Uniform scale of the foreground, 0.01–10.0, default 1.0 (100%).

​
rotation

Rotation of the foreground in degrees, -360–360, step 0.5, default 0.

​
keyframes

Advanced: a JSON array [{t, x, y, scale, rotation, opacity, interp}] that animates the transform over time. Normally driven by the node’s on-canvas keyframe UI rather than typed by hand.

​
background / foreground (inputs)

The two COMFYTV_VIDEO layers. background sits behind, foreground is blended on top. Both required.

​
mask (input, optional)

A COMFYTV_VIDEO mask that restricts where the foreground is visible.

​
track (input, optional)

A COMFYTV_TEXT motion/transform track (e.g. from a tracker). When connected, its keyframes drive the foreground transform, overriding the manual keyframes.

​
Outputs

​
Tips

  • The foreground is not chroma-keyed here — key or cut it out upstream (or feed a mask) so only the intended pixels merge.
  • HSL operators (hue, saturation, color, luminosity) and Porter-Duff ops (in, out, atop, xor, stencil, etc.) behave differently from the everyday over/screen/multiply — pick deliberately.
  • Video Transform — the same 2D transform on a single clip, without a second layer.
  • Video Transition / Luma Wipe — blend two clips over time instead of stacking them.