vlo Memory Load Video Batch

vloMemoryLoadVideoBatch is the busiest of the ComfyUI-vlo memory loaders, and the one that shows what this family is actually for. It takes an ordered collection of videos from an in-memory registry (fed by the vlo app, referenced by id, never dumped into ComfyUI/input) and hands them downstream as a Comfy video list - each clip an independent item, so durations and resolutions don't need to agree. Then, because we live in the era of models that generate sound with the picture, it carries a per-video audio decision alongside each clip.

The inputs

• files - the ordered multi-selection. A web extension swaps ComfyUI's static multi-select for an ordered selector that refreshes live from vlo's registry; arrow controls fix the exact order sent downstream. Capped at 100 items.
• disable_in_memory - flip it true and the selector loads from your ordinary input directory instead of the registry. This is how you test a workflow by hand without vlo running.
• include_audio - a comma-separated flag list in selection order, e.g. 1,0,1, one flag per video. The selector renders it as a checkbox on each row and vlo writes it from the speaker toggles on its batch slot. Unset items default to false, and the flags travel with their videos through adds, removals, and reordering - not positionally bolted on.

What comes out

Two ordered list outputs, guaranteed same order:

• videos - the VIDEO list itself.
• use audio - a BOOLEAN list, one flag per video.

That second output is the interesting bit, because it exists to be plugged straight into something. The MiniMax H3 reference adapter in this same pack (vlo MiniMax H3 Reference to Video (Batch)) takes a use_embedded_video_audio input that accepts a BOOLEAN list - so you wire the loader's "use audio" output to it and each reference video's soundtrack inclusion is decided per clip, not once for the whole batch. When MiniMax H3 treats a reference video's own sound as a separate <Audio N> reference that has to be explicitly enabled, having a per-video switch per batch is the difference between a workflow and a headache.

Install and the usual caveats

Whole pack, no extra deps:

cd ComfyUI/custom_nodes
git clone https://github.com/PxTicks/ComfyUI-vlo.git

then restart (or use ComfyUI Manager and search "ComfyUI-vlo"). If you're here without the vlo app, remember disable_in_memory - otherwise the node validates ids against the registry and rejects selections it doesn't know. The pack targets a recent ComfyUI, and if the audio inputs on your H3 node look different, that's a version gap on the ComfyUI side, not this node.