Nodes/Image Processing Suite for ComfyUI/Get First Frame (Batch to Single)
ComfyUI Node

Get First Frame (Batch to Single)

Grab frame zero out of a batch

By marcoc2·Created 2 years ago·Updated 5 months ago· 1
Get First Frame (Batch to Single)
  • images
  • IMAGE

Get First Frame (Batch to Single) is the smallest node in this pack and it knows it: it takes an IMAGE batch and returns images[0:1] - the first frame, as a single-image tensor. One line of logic, one output. That's the whole thing, and honestly, a node this trivial only survives in a pack because the job comes up so often that people get tired of writing it themselves.

You'll reach for it in the places where a pipeline wants one image and you're holding many. First-frame conditioning is the classic: image-to-video models like LTX are frequently set up with a "first frame" input, and if your source is a loaded video batch or a frame sequence, you need exactly the first one before anything else happens. Thumbnails, previews, "show me frame 0 of this sequence" - same deal. It's also the natural companion to the pack's own Walking Pose Generator, whose output is a batch of skeleton frames; grab frame one to preview or to feed a single-frame conditioning slot.

Mechanically there is nothing to say beyond the slicing: input images is a standard IMAGE, output IMAGE is that first frame with the batch dimension preserved as 1, so it plugs straight into any node that expects a normal image tensor. No parameters, no options, no way to pick a different index - if you want frame 5 instead of frame 0, this isn't your node, and that's by design.

The only judgment call is whether you even need it. Modern ComfyUI has a native way to slice batches in several node packs, and if you're already using a batch-math or slice utility, adding this is redundant. But if you'd rather not pull in another pack for one slice, or you want a node that self-documents its intent in a shared workflow ("yes, this deliberately takes the first frame"), this is a clean, legible choice. Zero VRAM, zero CPU, zero dependency - it's literally a tensor index.

Install is the standard AnotherUtils route: ComfyUI Manager → search "AnotherUtils", or cd ComfyUI/custom_nodes && git clone https://github.com/marcoc2/ComfyUI-AnotherUtils, then restart. No requirements, no models, nothing to download. If you already have the pack, it's already sitting in your node menu waiting for the one workflow where you need frame zero and you're out of slicing nodes.

CategoryAnotherUtils/image

Inputs (1)

NameTypeDefaultDescription
imagesIMAGE

Outputs (1)

NameTypeDescription
IMAGEIMAGE