Gigachad KSampler

Gigachad KSampler is the pack's star, and the honest description is: a drop-in replacement for the stock KSampler that bolts RES4LYF's sampler zoo onto your existing workflow. If you've seen workflow shares name res_2m, beta57, or "bongmath" and had no idea where those came from - they come from RES4LYF, and this node is one of the easiest ways to actually use them without rebuilding your graph.

The one-line version

It mirrors ComfyUI's own common_ksampler path - same model/positive/negative/latent/seed/steps/cfg inputs, same outputs. The difference is a clown_sampler dropdown that lists 119 RES4LYF sampler names (everything from multistep/res_2m to the fully implicit solvers), plus a scheduler list that includes beta57, plus a bongmath toggle. When clown_sampler is set to anything but none, it overrides the regular sampler - that's the intended behavior, not a bug.

The dependency you need to know first

The 119-name clown list only actually works if RES4LYF is installed. The code tries from custom_nodes.RES4LYF.beta.samplers import get_clown_sampler; if that import fails, it falls back to ComfyUI's own sampler lookup, which doesn't know res_2m from a hole in the ground. So plan on installing the prerequisite:

cd ComfyUI/custom_nodes
git clone https://github.com/ClownsharkBatwing/RES4LYF.git

restart, and only then start picking clown samplers. The regular sampler dropdown (44 stock names) and the standard schedulers work fine without it.

The settings worth understanding

• clown_sampler - leave on none unless you have RES4LYF installed. Then this is where the flow-matching magic lives: for a straighter-trajectory model, names like res_2m or res_2s with the beta57 scheduler are the community's default recommendation. The KB's sampler research calls beta57 "the single highest-value thing to try first" for flow-matching models like Anima, Wan, or Z-Image.
• bongmath - RES4LYF's "high-precision denoising" mode, aimed at Flux/DiT. It passes extra numerical parameters (via transformer_options) for improved accuracy. Per the tooltip, leave it off for SD1.5/SDXL - it's specifically a Flux/DiT thing.
• bongmath_cfg_scale (default 1) and bongmath_scale (default 0.5) - the CFG and noise scale used inside bongmath, independent of the outer cfg.
• scheduler - includes beta57 and linear_quadratic alongside the usual suspects. For flow-matching models, Karras is the wrong default (the KB calls it out as a distortion on straight trajectories); beta57 is where the action is.
• sigmas (optional) - override the scheduler with a custom SIGMAS schedule.
• options (optional) - a RES4LYF OPTIONS block for advanced ClownSampler control.

Outputs are latent (the normal result) and denoised_output (a cleaned version you can chain into further processing).

Install and the honest verdict

The pack itself installs like the rest of the family:

cd ComfyUI/custom_nodes
git clone https://github.com/Winnougan/comfyui-gigachad.git

or via ComfyUI Manager (comfyui-gigachad), then restart. No requirements.txt ships in the pack - but remember the RES4LYF prerequisite above if you want the headline feature.

Is this the best way to use RES4LYF? Not necessarily - installing RES4LYF gives you its own sampler nodes with the same access. But if you want a single familiar KSampler-shaped node that can do standard and clown samplers side by side, this is convenient and it works. It's the node in this pack with the most real capability behind the branding.