Load Scheduler Name

The companion to Load Sampler Name, and it works exactly the same way: it shows you ComfyUI's scheduler dropdown, and whichever one you pick comes back out as a plain STRING. Nothing gets scheduled, no noise gets removed, no model is even referenced. It's a label-picker, not a sampling step.

Why the exact string matters here more than you'd think

A scheduler decides how much noise gets removed at each step - a linear schedule removes equal amounts every step, while something like Karras front-loads the work into the middle steps. That distinction isn't cosmetic: on the flow-matching architectures that dominate 2026 (Z-Image, Qwen-Image, Wan, and friends), a Karras or Exponential schedule can actively hurt quality instead of helping, because those models expect a straighter, more balanced sigma curve than the older diffusion trajectories Karras was designed to correct. beta and simple tend to be the safer defaults there.

Which means if you're building a scheduler-comparison grid - the actual reason this node exists, per the pack's README, which frames its nodes as secondary helpers for qq-nodes-comfyui's XY Plot workflows - getting the exact right string into each column matters. Typing exponentiel instead of exponential fails silently or falls back to a default, and you've quietly wasted a row of your grid. Picking from the dropdown removes that risk entirely.

How it works

The scheduler widget lists ComfyUI's built-in schedulers: simple, sgm_uniform, karras, exponential, ddim_uniform, beta, normal, linear_quadratic, and kl_optimal - nine choices, matching what you'd see in a stock KSampler. The node's tooltip on the field, straight from the author, is "The name of the sampler's scheduler." Selecting one just converts it to text.

What to set, what it outputs

One field: scheduler, a nine-option dropdown. The single output is a STRING - tooltip "The scheduler name" - ready to feed a grid axis, a filename, or any node downstream that wants the scheduler identified as text rather than applied.

Installing it

ComfyUI Manager: search ComfyUI-ReplenishNodes, install, restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/hben35096/ComfyUI-ReplenishNodes

No model downloads, no extra Python packages documented in the README - it reads a list that's already built into ComfyUI.

Where people get tripped up

Same trap as its sampler sibling: the dropdown looks native, so it's tempting to wire the output straight into a KSampler's scheduler input and wonder why the connection is refused. The output is text, not a live scheduler object, and ComfyUI's type checking enforces that distinction for you.

Worth remembering if you go on to actually use whatever scheduler you picked: some third-party schedulers you'll see recommended for newer models - RES4LYF's bong_tangent is the well-known example - aren't in this node's list at all, because they ship inside their own node packs rather than ComfyUI core. This node only knows the nine built-in options; if your workflow leans on something more exotic, you'll need that pack's own selector, not this one.