🎬 Studio42 Video PatchDrop

Studio42 Video PatchDrop is the frame-by-frame version of the pack's PatchDrop: it places a patch video (or a single patch repeated) onto an original video at fixed coordinates, across every frame - with the temporal controls that keep the result from flickering like a bad green screen composite. It's the drop half of the Video PatchLift → PatchDrop pipeline, and it exists because dropping a patch per-frame manually is miserable.

Feed it the original video and a patch (same dimensions, same frame count - the typical pairing is the outputs of Video PatchLift Loader), and it composites frame by frame with masks, blend modes, feathering, scaling, and rotation. The video-specific extras are where it gets useful:

• temporal_consistency - smooths patch placement across frames so the patch doesn't jitter. The pack's docs suggest 0.3 as a starting value.
• motion_blur - simulates motion blur on the moving patch for natural integration.
• edge_blending - blends patch edges frame to frame so seams don't flash.

Everything else mirrors the still-image PatchDrop: patch_x/y, patch_mask, patch_width/height, patch_shape (+ corner_radius), blend_mode (9 modes), opacity, feather_edges, scale, rotation, position_mode (use_coordinates, center, manual with manual_x/y), resize_patch, preserve_aspect.

What comes out

result_video (the composited frame batch), placement_masks (per-frame masks of where the patch landed), and preview_video (with overlay indicators). Wire the result into a video save node with the audio from the loader and you're done.

Where it works - and where it doesn't

Fixed-coordinate placement across frames is great for a stable camera or a patch region that doesn't move. For a moving subject, the patch will sit still while the footage moves under it - you'd want per-frame coordinates upstream (or the pack's temporal controls set low and a stationary region). Within its lane it's a genuinely useful compositing node, and temporal_consistency + motion_blur are the knobs that keep it from looking like clip-art.

Installing it

Same pack, 24oiduts-ComfyUI:

• ComfyUI Manager: search 24oiduts, install, restart.
• Manual:

  cd ComfyUI/custom_nodes/
  git clone https://github.com/GeekyGhost/24oiduts-ComfyUI
  pip install -r 24oiduts-ComfyUI/requirements.txt

  then restart.

Like the other video nodes it needs opencv-python and ffmpeg-python (plus system FFmpeg) - if the whole Studio42 video section is missing from your menu, that's the cause. The GitHub README is a stale Studio42 template; ignore its "WIP, no license" banner.

Troubleshooting

• Flickering patch - raise temporal_consistency a bit, or make sure the region matches between lift and drop.
• Patch won't align - confirm both inputs have the same frame count and dimensions; the loader's frame_load_cap/select_every_nth outputs must match what you feed here.
• Hard seams - add feather_edges and turn on edge_blending.
• Video nodes missing entirely - dependencies; see install above.

For frame-consistent patching inside the graph, this is the node that finishes the job the Video PatchLift Loader starts.