IMAGE_EX Context (List)

IMAGE_EX Context (List) is the batch version of Ovum's image-context helper. It takes a list of IMAGE_EX dicts - the richer image record that the pack's "Load Image with Workflow" nodes produce, bundling the image, its mask, its filepath, and its embedded prompt/workflow metadata - and lets you override any of those fields per item, while the untouched parts pass through untouched.

The whole point is that you don't demux and remux. An IMAGE_EX is a bundle: {image, mask, filepath, prompt, workflow}. Without a context node, changing one field means unpacking the bundle into separate wires, editing, and rebuilding - and doing that for every item in a list. This node lets you send a list of IMAGE_EX in, provide list-form overrides, and get a list of patched contexts back, with everything else preserved automatically.

How it works

• All inputs are optional lists. Any input you leave unconnected is simply carried through from the base context.
• All provided list inputs must be the same length - the node enforces this and raises a clear error if you feed a 3-item image list but a 5-item filepath list.
• It works per-index: item i of the output is built from item i of each provided list. Safe copies are made, so your originals aren't mutated.

Inputs and outputs

• image_ex (LIST) - a list of base IMAGE_EX dicts (typically from Load Image with Workflow or Image List Loader).
• image, mask, filepath, prompt, workflow, edit_list - all LIST; per-item overrides. Note edit_list is threaded through but not merged into the context the way the others are.
• Outputs: IMAGE_EX, IMAGE, MASK, FILEPATH, PROMPT, WORKFLOW, EDIT_LIST - each a LIST. So you can grab just the list of patched images, or just the filepaths, or keep the full contexts.

That output structure is the practical win: want a list of images only? There's a socket for that, no unpacking needed.

When you'd reach for it

• Batch metadata round-tripping. Load a batch of images with their embedded workflows, patch the filepaths or prompts, pass the contexts on.
• Overriding masks in bulk. Feed a list of replacement masks alongside the image list; everything else stays intact.
• Stitching to the list variant of anything. Any node in the pack that wants list-shaped image data can be fed directly from these outputs.

Installing it

It's in sfinktah/comfy-ovum:

cd ComfyUI/custom_nodes
git clone https://github.com/sfinktah/comfy-ovum

or search comfy-ovum in ComfyUI Manager, then restart. The pack's requirements are small pure-Python helpers; no model downloads.

Gotchas

The length-matching rule is the one that bites: mismatch a single list and the whole node errors loudly rather than guessing. And remember IMAGE_EX is the pack's own convention - you'll mostly meet it coming out of Ovum's workflow-aware loaders, not from core ComfyUI's plain Load Image. If you're not already working with the pack's loaders, the single-image IMAGE_EX Context node is the gentler place to start; this list variant is for when you're processing a batch.