Image Conveyor

Good news: you don't have to learn anything new here. SequentialBatchImageLoader is not a separate node - it's the same Python class as Image Conveyor, registered under two names so older workflows and muscle-memory searches still resolve. Open the node picker and you'll find it under "Image Conveyor" either way. The pack's own search aliases - "sequential image loader", "batch image loader", "image queue" - hint at why the old name existed: it described the mechanics before the author settled on the conveyor branding.

So the behavior is identical. You drop images (or folders) onto the node, they show up as thumbnails in a visible in-graph queue, and the node hands them to your workflow one per prompt execution in queue order. Each item is tracked as pending, queued, or processed, with per-item controls to reset, force-done, or delete, plus bulk actions and manual drag-and-drop reordering. If you queue several prompt runs, the next items get reserved automatically - that's what the queue_item_json input is for, passing reservations between runs. And the optional Auto queue all pending toggle turns a single queue action into one queued prompt per remaining image.

Outputs

Same schema as the main node: image and mask (the loaded image plus its alpha, if any), path (the ComfyUI-side annotated input path actually loaded), index (1-based queue position), remaining_pending (how many are left after this run), and source_path (a best-effort original drop path - filename-only for absolute native paths, deliberately, so exported workflows don't leak your local directories).

Install

The pack is xmarre/ComfyUI-Image-Conveyor, installable from ComfyUI-Manager as "ComfyUI Image Conveyor" or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/xmarre/ComfyUI-Image-Conveyor.git

Restart after cloning. There are no model downloads and the package declares zero Python dependencies - the only real requirement is a modern ComfyUI frontend, since the gallery renders through the supported custom-widget/DOMWidget path (VueNodes-compatible).

When you'll actually see this name

Mostly in downloaded or older workflows. Anyone who saved a graph with this node before a rename (or grabbed one from a post) will have SequentialBatchImageLoader in the JSON, and it just loads. If you're building fresh, use Image Conveyor - same node, same inputs (state_json, ui_state_json, queue_item_json are internal plumbing, not things you set by hand). One genuinely useful quirk to remember: if a run fails before the loader executes, items can stay stuck on queued, and the Clear queued action is how you release them.