FL Image Batch To Image List

ComfyUI has two different ways of carrying "more than one image" through a graph: a batch (one tensor, stacked along a batch dimension, everything the same size, processed in parallel) and a list (individual items, processed one at a time, allowed to differ). Most sampling and generation nodes hand you a batch. Plenty of downstream tools - anything doing per-item work, some captioning nodes, some export nodes - want a list instead. FL_ImageBatchToImageList is the node that gets you from one to the other.

What it does is genuinely that simple. Feed it image - a batched tensor, straight from a sampler, FL_ImageRandomizer's image_batch output, wherever - and it iterates through it index by index, handing back the same images as a Python list of single-image tensors, one item per batch slot. There's no processing, no transformation of the pixels themselves; it's purely a container-shape change.

It's the direct inverse of FL_ImageListToImageBatch, which is the same pack's node for going the other direction - bundling a list back into a batch when a downstream node wants that shape instead.

Why you actually run into this. Batches are the efficient shape for anything that treats all your images identically - a sampler, a resize, a single blend operation applied uniformly. But the moment you need per-image logic - saving each image to a differently-named file, running a per-item caption node, feeding images one at a time into something like FL_GPT_Image1 or FL_SimpleGPTVision where each call is its own API request - you need a list, because a batch operation processes everything as one parallel unit and has no concept of "do something slightly different for item 3." This node is the seam between those two worlds: generate in batch for speed, then split to a list the moment you need to treat each image as its own thing.

Installing it is the standard Fill-Nodes routine: search "Fill-Nodes" in ComfyUI Manager, or clone it manually -

cd ComfyUI/custom_nodes
git clone https://github.com/filliptm/ComfyUI_Fill-Nodes

• then restart ComfyUI. No dependencies beyond what the base pack already needs; this is pure Python iteration over a tensor.

Troubleshooting. There's really only one class of problem with a node this minimal, and it's directional: if a downstream node errors out complaining it wanted a batch and got something else (or vice versa), you've likely gone through the wrong converter, or you needed one and skipped it entirely. If you're not sure which shape a given node in your graph expects, the practical test is just trying it - ComfyUI's type system generally surfaces a batch/list mismatch as a clear error rather than a silent wrong result, so when you see that error, this node (or its inverse, FL_ImageListToImageBatch) is almost always the fix.