采样器名称

LamSamplerName (采样器名称, "sampler name") is a pass-through selector: a dropdown of every sampler ComfyUI knows, and an output that hands the chosen name to whatever you connect it to. That's it. No math, no sampling - it exists to solve a genuinely annoying workflow problem.

The problem: once a workflow grows to three or four KSampler nodes (base pass, hires fix, a detail pass), changing the sampler means clicking into every single sampler and re-picking. Miss one and your "uniform" change silently isn't uniform. LamSamplerName centralizes the choice - connect its output to each KSampler's sampler_name input, and one dropdown controls them all. Flip the workflow from euler_ancestral to dpmpp_2m and every sampler follows. It's the same trick the scheduler-sibling LamScheduler does for schedulers, and the same QoL idea that made rgthree's nodes a staple.

How it works

The dropdown is populated directly from ComfyUI's own sampler registry - comfy.samplers.KSampler.SAMPLERS - so it's not a hardcoded list that drifts out of date when a new sampler lands. There are 44 entries in current builds, from the everyday (euler, euler_ancestral, dpmpp_2m) to the niche (exp_heun_2_x0_sde). The author's tooltip is honest about what the field does: "The algorithm used when sampling, this can affect the quality, speed, and style of the generated output." The node takes that string and emits it unchanged.

Because the output is typed as * (wildcard), it plugs into any node with a sampler-name input - KSampler, the pack's own samplers, or third-party sampler wrappers that take a string. It won't coerce into a hard STRING-typed input if the target node is strict about it, but the standard KSampler family accepts it fine.

Inputs and output

• sampler_name - the dropdown.
• Output: sampler_name - the chosen string, ready to wire into every KSampler in the graph.

Install

In yanlang0123/ComfyUI_Lam - Manager search "ComfyUI_Lam", or:

cd ComfyUI/custom_nodes
git clone https://github.com/yanlang0123/ComfyUI_Lam

then restart. Zero dependencies beyond the pack import, no models - the README's install.bat requirement list is irrelevant to this node.

Gotchas

The wiring is the gotcha: a single LamSamplerName output can drive multiple samplers, but each connection is one wire, and if you later replace one of those samplers with a copy that doesn't have the wire yet, that new sampler falls back to its own default. Reconnect, don't assume. Also note it only carries the sampler name - it does nothing for the other per-sampler settings (steps, CFG, scheduler), so it's a one-knob convenience, not a workflow-wide settings hub. For a node with 44 options and one function, that's the whole deal - and it's a good deal when you're iterating on sampler choice across a big graph.