LC Final Frame πΌοΈ
Grab the last frame of a clip for a still, a thumbnail, or a loop-closer
- images
- image
When a video workflow finishes you often need a still - a poster frame, a cover image, a thumbnail for a video you just saved. LC Final Frame is the one-line answer: feed it the IMAGE batch (the frame sequence, playback order) and it returns the last frame as a single image.
It's the same contract as the FinalFrameSelector nodes from other packs, so this is a like-for-like drop-in if you're consolidating. And if you're thinking "why the last frame instead of the first?" - both are legit, they just serve different jobs. The first frame is your poster; the last frame is your checkpoint, the thing that tells you how the clip actually ended. Some workflows also use it to chain - take the final frame of one generated segment and feed it as the first-frame conditioning of the next, which is how a lot of multi-shot video gets stitched (the ecosystem docs cover how chaining on last frames is the standard extension trick). For that job, grabbing the true last frame automatically beats trusting your eyes on where the clip ends.
How it works
Nothing magical: an IMAGE batch is a tensor with shape [B, H, W, C], and this node returns images[n-1:n] - the final entry, keeping the batch dimension so it stays a valid IMAGE tensor. If you give it an empty batch it raises a clear error rather than silently returning garbage. That's the whole mechanism, which is why it's dependable: it's pure tensor indexing, instant, and correct for any batch of frames regardless of model.
The inputs and outputs
images- the frame batch. Last frame wins.- Output:
image- a single-frame IMAGE you can save, preview, upscale, or feed into image-to-video as your new starting frame.
Typical wiring: take the images output from an LC Create Video (or straight off a VAE decode) into this node, then out to a Save Image node or into the next video stage.
How to install it
Ships in lonecatone23/ComfyUI_LC_AV_nodes, the LC Audio_Video toolkit (MIT, by lonecatone23; sibling to the image-side ComfyUI_LC123_nodes - don't merge the two folders):
- ComfyUI Manager: search "LC Audio_Video" or
ComfyUI_LC_AV_nodes. - Manual:
cd ComfyUI/custom_nodes && git clone https://github.com/lonecatone23/ComfyUI_LC_AV_nodes, restart ComfyUI.
Console should print [LC AV] total 16 nodes. No dependencies beyond core and no downloads.
Gotchas
A frame is a still - the output has the same resolution as your video frames, so it's not a magical high-res render; upscale it separately if you need a big cover image. And remember the batch is in playback order, so "last" means last in sequence. If your pipeline ever reverses or shuffles frames upstream, the "final frame" you get is only as meaningful as the order you fed in - for a straight generation it's exactly the clip's last visible frame.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| images | IMAGE | Frame batch. Last frame is returned. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | β |