Load Images From String

Most image loaders in ComfyUI want you to pick a file from a dropdown or an upload widget. This one takes a different approach: you paste in a list of file paths, one per line, and it loads all of them into a single batch. It's a text box in, an image batch out.

That makes it a scripting-and-automation node more than a point-and-click one. If you've got an exact set of files you want - specific frames, a curated selection, output from some earlier step that wrote paths - you list them and load them in one shot. It shines in headless or programmatic setups where the paths are generated by something else and you're driving ComfyUI without touching the UI. For casually browsing your image folder, the standard loaders are friendlier; for "load these exact 40 files," this is the one.

How it works

It reads the multiline string, treats each line as a path, and loads each image, stacking them into one batch tensor. The default text in the widget (./frame000001.png and friends) is just an example of the one-path-per-line format.

Inputs and outputs

• paths - a multiline string, one file path per line. Relative paths resolve against ComfyUI's working directory.
• ignore_missing_images - false or true. Leave it false and a missing file is an error that stops the run; set it true and missing files are silently skipped, which is handy when your list might have gaps.

Output is a single IMAGE batch containing everything it loaded.

Installing it

Found in the comfyui_image_ops file. Install through ComfyUI Manager (search Various ComfyUI Nodes by Type / comfyui-various) or clone:

cd ComfyUI/custom_nodes
git clone https://github.com/jamesWalker55/comfyui-various

Restart ComfyUI. No models or extra dependencies for this node.

Common issues

The one that catches people: the images have to be the same size to stack into a batch. ComfyUI batches are a single tensor, so every image in it must share dimensions. List a mix of resolutions and the load will fail. If your files aren't uniform, you'll need to resize them to a common size first (or load and resize individually).

Second, paths. Relative paths are resolved against wherever ComfyUI is running from, not against wherever your workflow file lives, so a path that "looks right" can still miss. When in doubt, use absolute paths. And flip on ignore_missing_images if a stray missing file shouldn't nuke the whole run.

One more thing that matters on a hosted or serverless runner like this platform: the filesystem is ephemeral and you don't control it the way you do on a local install, so hardcoded local paths from your own machine won't exist there. This node is at its best on a local ComfyUI where you own the disk; in the cloud you'd typically bring images in through the normal upload path instead.

If it shows up red as a missing node in someone else's workflow, that's just the pack not being installed yet - Manager's "Install Missing Custom Nodes" or a clone of the repo fixes it.