split

When a node has a one-word display name - just "split" - you can usually guess the whole deal. Give it a path like /output/frames/scene_003.png, and it hands back the directory (/output/frames) and the filename (scene_003.png) as two separate strings. That's the entire feature, and it's the load-bearing step in a lot of "save next to the input" workflows.

The reason you want it broken apart is that once directory and filename are separate strings you can rebuild paths flexibly: keep the same folder but swap the file, or keep the file and point it at a different directory. String nodes can do this with split on /, but that's fragile - Windows separators, trailing slashes, and drive letters all break naive string splitting. This uses os.path.split(), which is platform-aware and knows how to handle all of it. Same result, no footguns.

How it works

It's a straight passthrough to os.path.split(). Worth internalizing two Python behaviors so you're not surprised:

• A trailing slash means the "filename" part is empty ("/a/b/" → ("/a/b", "")).
• A bare filename with no directory returns an empty directory string ("scene_003.png" → ("", "scene_003.png")).

Both are correct os.path behavior; neither is a bug, but they explain a few "why is this string empty" moments.

Inputs and outputs

One input, path (default empty string). Two outputs:

• directory - everything up to the last separator.
• filename - the last path component.

Wire directory into a PathJoin or a save node, and filename into a PathSplitExt if you then want the stem without the extension. The two nodes compose nicely: split → splitext is the standard "folder + base name" pipeline.

Installing it

Part of the Basic data handling pack by StableLlama:

cd ComfyUI/custom_nodes
git clone https://github.com/StableLlama/ComfyUI-basic_data_handling

Restart ComfyUI, or grab it via ComfyUI Manager (search "Basic data handling"). Zero dependencies, no downloads - installing it can't conflict with anything, which is the pack's whole pitch.

Gotchas

• It's pure string work: the path doesn't need to exist, and the node won't check.
• Don't confuse this with PathSplitExt (the "splitext" node) - that one splits the extension off the filename. Split splits at the last slash; splitext splits at the last dot. Different axes, and you'll reach for both within the same workflow.