🔶 Any input2array

This one does exactly what the name says and nothing more: plug in up to four things, get one ARRAY out. No conversion, no math, no type checking - it's a collector. If you've used a "combine" or "bundle" node in another pack, same idea, just wired to Chaosaiart's own ARRAY type instead of a list.

Where it actually earns a slot in your graph is Chaosaiart's frame-by-frame animation setup. That whole pack is built around generating one image per "frame" (their word for one run through the queue, not a video-timeline frame) and changing checkpoints, prompts, or ControlNet settings as the count climbs. Once you're juggling several of those changing values, an array is a tidier way to pass a group of them around than four separate wires crossing your whole canvas. The pack's own README calls this out directly: use arrays "to enhance the organization of your workspace, or swiftly switch between processes to test things" - so beyond tidiness, it's also handy for quickly A/B-ing a batch of checkpoints or prompts by swapping what's plugged into slot 1 through 4 rather than rewiring.

Inputs and outputs

• input_1 - required. This is the only slot you have to fill; leave 2–4 empty and you get a one-element array.
• input_2, input_3, input_4 - optional, same wildcard * type as input_1. Fill as many as you need.
• ARRAY - the only output. Wire it into whatever downstream node in the pack expects an array (or into your own logic if you're scripting against it).

Since every input is typed *, ComfyUI won't stop you from mixing types - you could technically feed it a checkpoint on input_1 and a string on input_2. Nothing enforces that they're related, so that's on you to keep straight.

Installing it

Easiest path is ComfyUI Manager: open it, choose Install via Git URL, and paste https://github.com/chaosaiart/Chaosaiart-Nodes - that's the exact method the pack's own README documents, rather than a name search. If a title search for "Chaosaiart" turns it up for you, that works too.

Manual install:

cd ComfyUI/custom_nodes
git clone https://github.com/chaosaiart/Chaosaiart-Nodes

On Linux (and inside a venv, activate it first):

pip install opencv-python
pip install tqdm

Windows users get an Install_windows script bundled in the node folder that handles the same two packages. Restart ComfyUI after either path. No models to download - this is a pure logic pack, it just routes whatever you already have loaded.

Common issues

The whole pack fails to load, not just this node. Every Chaosaiart node lives in one Python module, so if opencv-python or tqdm didn't install cleanly, ComfyUI's console will show an import error on startup and you'll lose access to all ~20 nodes, not just this one. Re-run the pip installs (inside your venv if you use one) and restart.

No tooltips, no node description. This pack shipped as a solo project aimed at solving the author's own animation workflow first - the in-app descriptions are blank across the board. The GitHub README's workflow JSONs are the actual documentation; if you're unsure how a node's meant to connect, download one of those example workflows and look at the wiring rather than guessing from the node alone.

Don't expect it to validate anything. Since the inputs are wildcard-typed, a bad connection (wrong data landing in the array) won't throw an error here - it'll surface downstream, wherever you actually try to use the array's contents. If something breaks further down the chain, check what actually went into this node first.