LC Sampler Configure Simple (pipe)
One CFG, packed into the bundle
- pipe
- pipe
- total_steps
- cfg
- denoise
- sampler_name
- scheduler
- detailer_steps
This is the pipe-flavored version of LC Sampler Configure Simple, and it has a nice trick up its sleeve: it speaks both pipe-world and single-CFG world. You get one CFG, steps, denoise, sampler, and scheduler - but instead of just spitting out sockets, it also packs everything into an LC_PIPE that plays nice with the dual-pass machinery.
The headline behavior is in the author's own description: when it packs the pipe, it sets step_swap = 0 and cfg_2 = cfg. What that means in practice: a consumer that expects the full dual-pass key set (like the pipe-fed sampler stack) sees a "swap at step zero, second CFG equals the first" schedule, which effectively degrades to the single-pass behavior you asked for. You don't have to know whether your downstream graph wants the dual-pass shape - this node satisfies both.
How to use it
- pipe (optional) - an incoming
LC_PIPE. This node's sampler keys overwrite matching keys; every other field passes through. Leave it empty and it builds a fresh bundle from the widgets. - Widgets: total_steps, detailer_steps, denoise, cfg, sampler_name, scheduler - same dials as the non-pipe Simple variant.
Outputs: pipe, then total_steps, cfg, denoise, sampler_name, scheduler, detailer_steps. So you can ride the pipe onward for the rest of the graph and still wire the plain values straight into a KSampler if that's more convenient at the edge.
Install
Same pack as the rest of LC123 - Manager (search "LC123") or:
cd ComfyUI/custom_nodes
git clone https://github.com/lonecatone23/ComfyUI_LC123_nodes
Restart. No extra pip packages (ComfyUI's torch/numpy suffice), MIT licensed. On cloud-hosted ComfyUI, LC123 must be requested or run locally.
The take
It's a nice compatibility bridge rather than a must-have: if your graph is all pipes, this is the single-CFG config node that doesn't force you to abandon the bundle. If your graph is all explicit wires, the plain Simple version is cleaner - fewer sockets to trip over, and no LC_PIPE type hanging in the middle of it. Pick based on whether the pipe is already your pattern.
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| total_steps | INT | 401–10000 | Total sampling steps. |
| detailer_steps | INT | 00–10000 | Steps reserved for a detailer / refiner stage (0 = unused). |
| denoise | FLOAT | 1.000–1 | Denoise strength (1.0 = full). |
| _gap1 | STRING | Layout spacer | |
| cfg | FLOAT | 8.00–100 | Classifier-free guidance scale. |
| _gap2 | STRING | Layout spacer | |
| sampler_name | COMBO | euler | Sampler algorithm. |
| scheduler | COMBO | normal | Noise schedule. |
| pipeopt | LC_PIPE | Optional LC_PIPE in. Sampler keys from this node overwrite; other keys pass through. |
Outputs (7)
| Name | Type | Description |
|---|---|---|
| pipe | LC_PIPE | — |
| total_steps | INT | — |
| cfg | FLOAT | — |
| denoise | FLOAT | — |
| sampler_name | euler,euler_cfg_pp,euler_ancestral,euler_ancestral_cfg_pp,heun,heunpp2,exp_heun_2_x0,exp_heun_2_x0_sde,dpm_2,dpm_2_ancestral,lms,dpm_fast,dpm_adaptive,dpmpp_2s_ancestral,dpmpp_2s_ancestral_cfg_pp,dpmpp_sde,dpmpp_sde_gpu,dpmpp_2m,dpmpp_2m_cfg_pp,dpmpp_2m_sde,dpmpp_2m_sde_gpu,dpmpp_2m_sde_heun,dpmpp_2m_sde_heun_gpu,dpmpp_3m_sde,dpmpp_3m_sde_gpu,ddpm,lcm,ipndm,ipndm_v,deis,res_multistep,res_multistep_cfg_pp,res_multistep_ancestral,res_multistep_ancestral_cfg_pp,gradient_estimation,gradient_estimation_cfg_pp,er_sde,seeds_2,seeds_3,sa_solver,sa_solver_pece,ddim,uni_pc,uni_pc_bh2 | — |
| scheduler | simple,sgm_uniform,karras,exponential,ddim_uniform,beta,normal,linear_quadratic,kl_optimal | — |
| detailer_steps | INT | — |