Images Range

Your video loader spat out 200 frames and you only want frames 50 through 100. Or your batch generator produced a long strip and you want to drop the first few and the last few before processing the rest. Images Range is the slicer: it cuts an image batch down to a sub-range, with optional start and end boundaries that you can switch on independently.

The mechanism is a slice, and the source is refreshingly honest about it: the node literally does images[start:end] - with a twist. Each boundary has a boolean switch: use_start gates whether start is applied, use_end gates whether end is applied. If a switch is off, that bound is None, meaning "don't trim this side." That's why the defaults work the way they do: use_start and use_end both default to false, so out of the box the node is a pass-through - it returns the whole batch unchanged. You opt into trimming per side.

The inputs:

• images - the batch to slice.
• start - the start index, applied only if use_start is on.
• use_start - switch for the start bound.
• end - the end index, applied only if use_end is on.
• use_end - switch for the end bound.

The output:

• images - the sliced batch.

Because this is a Python slice, all the usual semantics apply: end is exclusive (so start=50, end=100 gives frames 50 through 99), and negative values count from the end of the batch. Need to drop the first 10 frames of a 200-frame batch? Set start=10 and flip use_start on. Need the last 30? Leave start off and set end=-30 with use_end on - negative end keeps everything except the final 30.

Where it fits: any batch-trimming need - cutting dead frames off a video sequence before upscaling, isolating a middle segment of a batch for processing, or pulling a sub-range out to feed a different node than the full batch. It's the "many" counterpart to the pack's Images Index, which grabs a single frame.

Honest gotchas: slicing doesn't validate your bounds, so start beyond the batch length yields an empty tensor rather than an error - downstream nodes may then fail confusingly, so keep your indices sane. And both switches off is a silent pass-through, which is easy to mistake for a bug if you set values and forget to flip the switches. The node does exactly what the source says: nothing until you tell it to.

Install is the standard pack path: ComfyUI Manager → search ComfyUI-FairLab → install → restart:

cd ComfyUI/custom_nodes
git clone https://github.com/yanhuifair/ComfyUI-FairLab.git
cd ComfyUI-FairLab
pip install -r requirements.txt

Restart, search "Images Range" or "batch slice". No models, no dependencies - just a slice.