Create Int List

This node exists for exactly one job: building the t_index_list that StreamDiffusion Create Stream needs and that ComfyUI has no native widget for. There's no generic "list of integers" input type in stock ComfyUI, so the pack ships its own tiny builder - punch in how many numbers you want, fill in the values, get a LIST out the other end.

If you're not touching StreamDiffusion, you'll never need this node. Its entire reason to exist is that one downstream connection.

What t_index_list actually is

StreamDiffusion's real-time trick is skipping most of the denoising trajectory and only stepping through a handful of pre-selected timesteps - that's what makes it fast enough to run near-live instead of taking the usual 20-50 steps per image. t_index_list is that handful: a short list of integer indices into the scheduler's step sequence, telling StreamDiffusion exactly which points along the noise schedule to actually compute. Fewer entries means fewer steps means faster generation, at the usual cost of a distilled or LCM-paired pipeline: less room for the model to correct itself, so the base checkpoint and (typically) an LCM-LoRA need to already be tuned for very-few-step generation before this shortcut works well.

The inputs and outputs that matter

• elements_count (required, default 2, range 1–10) - how many of the element slots below are actually read. Bump this up before filling in more than two values, or the extras get ignored.
• element_1 through element_9 (optional, default 0 each) - the individual integers. Only the first elements_count of these get used.

One output: LIST, wired directly into StreamDiffusion Create Stream's t_index_list input.

How to install it

ComfyUI Manager → Install Custom Nodes → search ComfyUI-Diffusers → install → restart, or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/Limitex/ComfyUI-Diffusers.git
cd ComfyUI-Diffusers
pip install -r requirements.txt

Since you're building this specifically to feed StreamDiffusion, you'll also want the rest of the StreamDiffusion setup from the README: clone cumulo-autumn/StreamDiffusion inside the node's folder and run python -m streamdiffusion.tools.install-tensorrt if you want the TensorRT-accelerated path. That step is heavier than the base pack install and worth doing deliberately rather than skipping past - it's the difference between StreamDiffusion running at a crawl and running close to real time.

Common issues & troubleshooting

"Missing node type: CreateIntListNode." Standard fix for this whole pack: install ComfyUI-Diffusers via Manager or git clone, then restart ComfyUI.

Set element_5 but nothing changed. Check elements_count first - it defaults to 2, so anything past element_2 is silently ignored until you raise that number to match how many values you actually filled in.

StreamDiffusion Create Stream errors on t_index_list. This is usually not a Create Int List problem but a values problem: t_index_list needs indices that make sense for the scheduler's step count configured elsewhere in your stream setup - an index past the end of the sequence, or an empty/zero-length list, will error out downstream rather than here. Cross-check against how many steps your stream is actually configured for.

Not sure what values to use. There's no single universal answer - it depends on your checkpoint, your LCM-LoRA (if any), and how aggressive a speed/quality tradeoff you want. Start from a working example workflow if you can find one rather than guessing values from scratch; this is one of the more fiddly settings in the whole StreamDiffusion setup and small changes here have an outsized effect on output quality.