Wan Wrapper Sampler Default JK🐉

Wan video generation in ComfyUI mostly happens through Kijai's Wan Video Wrapper, and the wrapper's sampler node is picky: it wants a specific scheduler string and a specific rope-function setting, and getting those wrong means slow runs, garbled motion, or outright errors. Wan Wrapper Sampler Default JK is JakeUpgrade's helper for that - a curated dropdown of the scheduler names the wrapper actually accepts, plus the matching rope-function value, so your workflow carries the right settings without you typing strings from a forum post into a text field.

How it works

The node has one input - scheduler - and a dropdown that's pre-filled with the wrapper's accepted values:

unipc, unipc/beta, dpm++, dpm++/beta, dpm++_sde, dpm++_sde/beta, euler, euler/beta, longcat_distill_euler, deis, lcm, lcm/beta, res_multistep, flowmatch_causvid, flowmatch_distill, flowmatch_pusa, multitalk, sa_ode_stable, rcm.

It returns that same scheduler string plus a rope_function string, which is hardcoded to "comfy" - the standard rope-prefill option for the Wan wrapper. You wire both into the wrapper's sampler settings.

The list is the giveaway to how Jake uses Wan: the longcat_distill_*, flowmatch_*, multitalk, and rcm entries exist because JakeUpgrade's Wan subgraphs cover specialized workflows - LongCat long-video, FlowMatch-based fun models, lip-sync multitalk, and so on. If you're on a plain Wan 2.1/2.2 run, unipc is the safe default that most of those workflows use.

Inputs and outputs

• scheduler - pick from the curated list.
• scheduler - the chosen value, passed through for the wrapper.
• rope_function - always "comfy", for the wrapper's rope setting.

Installing it

Ships in ComfyUI-JakeUpgrade:

• ComfyUI Manager → search "JakeUpgrade" → install → restart.
• Or:

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

Windows portable: run install.bat or ../../../python_embeded/python.exe -s -m pip install -r requirements.txt.

Note: the node itself doesn't load any Wan model. The actual weights come from the Wan Video Wrapper pack, which is a separate install and a big one - if you loaded a JakeUpgrade Wan workflow and it's complaining about missing nodes, the wrapper (or VideoHelperSuite, etc.) is usually the missing piece, not this node.

Common issues

• It's not a KSampler. This node produces settings strings, not a sampler. It feeds the Wan Wrapper's sampler node, which does the actual denoising. Wiring it into a plain KSampler won't work.
• A scheduler mismatch causes weird results, not always an error. The wrapper may silently accept an unfamiliar string and produce wrong motion or slower sampling. That's precisely why the curated dropdown exists - use it instead of typing.
• rope_function is fixed to "comfy". If you're on a wrapper version or workflow that expects a different rope function, this node won't give it to you. That's the trade-off of a "defaults" node: convenient for the standard path, rigid elsewhere.
• It assumes you're on the Wan Wrapper ecosystem. If you're using native Wan nodes instead of the wrapper, this helper doesn't apply.

For anyone running JakeUpgrade's Wan subgraphs, this is the node that makes "what scheduler do I feed the wrapper" a solved problem - one dropdown, always valid choices, zero string-typing.