ER Wedge Preview Video

When you wedge video, the deliverable problem gets worse than with stills. A client can look at four graded frames side by side and pick one. Four versions of a clip - where the difference is in the motion, the flicker, the temporal stability - need to actually play to be judged, and that's exactly what a thumbnail contact sheet can't give you.

ER Wedge Preview Video is the presentation layer for the ER Wedge Sequencer Video node. Where the sequencer gives you a functional grid for working, this is the clean gallery for showing: it mirrors every video your video-sequencer nodes produced, grouped per wedge and labelled with the parameter values, with each clip autoplaying in the gallery. It's the frame-batch version of ER Wedge Preview Image, from the same ER VFX Nodepack by Héctor Gallego (ER Academy).

No wiring, by design

Like its still-image sibling, this node has no inputs and no outputs - the Python side is a no-op and all the work happens in the front-end, which watches the graph for ER Wedge Sequencer Video results and mirrors them. You drop it anywhere and it fills itself in. Each sequencer's videos show up grouped (WEDGE 1, WEDGE 2...), labelled with the values that produced them, so a steps-vs-flicker test reads as "clip at 8 steps, clip at 12 steps" instead of a row of anonymous movies. Type a custom title for the session, and export a contact sheet - one PNG built from the first frame of each video, with its labels - via the 🖼 button or right-click.

Where it shines

• Client-facing look development. Send one PNG contact sheet, or sit them in front of the live gallery and let clips autoplay. This is the "here are your options" step of real VFX work, minus the part where you export a folder of videos and paste labels into a doc.
• Internal A/B memory. Since labels travel with the clips, you can come back to a wedge session next week and still know which settings made which motion artifact.

Honest limitations

It only mirrors results from ER Wedge Sequencer Video nodes in the current graph - it won't ingest arbitrary videos. It doesn't run wedges itself. And because the underlying clips live as MP4 proxies, the videos you see in the gallery are proxy quality (capped for smooth playback); clicking opens the full file, and a workflow run regenerates them. The sequencer's video side needs PyAV (pip install av if it isn't already bundled with your ComfyUI) - if that import fails, this node has nothing to mirror.

Installing

It's part of the ER VFX Nodepack, not a standalone download. In ComfyUI Manager search "ER VFX Nodepack" → Install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/esperandoelrender/ComfyUI-ER-VFX-Nodepack.git

Restart and find it under utils. If you previously installed the individual er_* packs this pack replaces, remove those folders first to avoid duplicate node registrations.