🪐 Load Videos From Folder (Simple)

Load Videos From Folder (Simple) does exactly what the name promises and nothing else: point it at a directory, and it loads every video inside, concatenates the frames into one big image batch, and merges the audio into a single track. If you're feeding a batch-joining workflow - a folder full of clips you want stitched into one sequence - this is the thing that stops you from wiring up fifteen Load Video nodes.

It's from the same author as the VACE Join nodes, and it exists to solve the "where do the clips come from" half of that pipeline. No extra dependencies, no ffmpeg CLI wrangling, no per-file nodes.

How it works

The loader decodes every supported file (webm, mp4, mkv, gif, mov) with PyAV, reads all frames into one tensor, and resamples each clip's audio to a common sample rate so the combined track is coherent. All videos have to share the same resolution - there's no resizing here, by design; the downstream VACE nodes require matching dimensions anyway, so it fails fast if you give it a mixed folder.

Two caveats worth knowing. First, the audio output is disabled (empty) if any video in the folder lacks an audio track - it doesn't try to fake it. Second, for large collections the all-at-once load can eat your system RAM. That's what the optional meta_batch input is for.

The inputs and outputs that matter

• folder_path - full path to the directory holding your videos. The only thing you must set.
• debug (default false) - logs per-file progress to the console. Turn it on the first time to see which files it picked up.
• meta_batch (optional) - connect a VideoHelperSuite Meta Batch Manager here to load large collections in RAM-safe chunks. VHS is a separate install, and only needed when concatenating genuinely long inputs.

Outputs: images (the concatenated IMAGE batch, ready for video creation) and audio (the combined track).

Install

ComfyUI Manager → search "Wan VACE Prep" → install → restart, or:

cd /path/to/comfyui/custom_nodes
git clone https://github.com/stuttlepress/ComfyUI-Wan-VACE-Prep

The README claims no external dependencies, and that's practically true - the code imports PyAV, but av ships as part of ComfyUI's own stack, so there's nothing extra to install. VideoHelperSuite is separate; you only add it if you need meta_batch.

Common issues

• "Folder does not exist" - paste the path carefully; the node strips surrounding quotes, but a typo is a typo.
• "No video files found" - the error lists the supported formats. Watch out for .mp4 vs .MP4 casing on Linux.
• Resolution mismatch - the loader checks and errors the moment a second video doesn't match the first. Standardize your clips (VHS Video Resize is the usual tool) before pointing this at a folder.
• Out of memory on a big batch - concatenating hundreds of clips in one tensor is exactly what causes this. That's the Meta Batch Manager use case: VHS's docs suggest roughly frames_per_batch = 10× your RAM in GB (32 GB → ~320 frames per batch).