Load Images (Video)

Pulls the frames out of a FACELESS_VIDEO and hands them back as a standard IMAGE batch - the bridge between this pack's own video world and the rest of ComfyUI. It's not a minor utility node; it's one of the three example workflows the README ships, literally titled "Preview video frames," and the README explains exactly why in the same breath: this pack doesn't support previewing video yet. This node is the workaround.

Why it matters more than it looks

None of this pack's video nodes have an inline preview - no scrubber, no thumbnail, nothing like the video-player widgets you might know from other packs' video-combine nodes. That makes this the only way to actually look at your video mid-pipeline without saving the whole thing and opening it in a media player. Drop one after any node that outputs FACELESS_VIDEO - a face swap, a restoration pass, a background removal - feed it into a PreviewImage node, and you get a cheap sanity check on what that step actually produced before you commit to rendering the rest of the pipeline.

Once you're back in IMAGE territory, everything opens up: PreviewImage to look at it, SaveImage to grab specific frames, or any image node from any other pack - none of which can touch a FACELESS_VIDEO directly. That's not just a preview trick, either - it's also the only way to run a non-Faceless node against your video's frames at all, whether that's a different pack's upscaler, an unrelated background remover you want to A/B against this pack's own, or a standard SaveImage to pull out one specific frame as a still.

Inputs and outputs

• video - the FACELESS_VIDEO to unpack.

Output: images, a standard IMAGE batch - every frame from the clip.

Installing it

Manager: search Faceless Node for ComfyUI. By hand:

cd ComfyUI/custom_nodes
git clone https://github.com/jeffy5/comfyui-faceless-node
pip install -r requirements.txt

No model needed for this node - it's a straight type conversion, not a machine-learning step.

Where people get burned

A long clip means a lot of images at once. This pulls every frame in the video into one batch - on anything beyond a few seconds, that's a large IMAGE batch landing in a single PreviewImage grid. For a quick sanity check, pair this with a trim on your loader (Load Video's or Load Video (Url)'s trim_frame_start/trim_frame_end) so you're only decoding and previewing a handful of frames rather than the whole clip.

It's a preview workaround, not a general video-editing tool. There's nothing here to select specific frames or downsample - it hands back everything the input FACELESS_VIDEO has. If you need a subset, trim upstream first.