Sigmas Pad

This is a small, deliberately dumb node, and that's the point: it takes a sigma schedule and adds one more entry onto it, set to whatever value you choose. No shape math, no rescaling, just a single value tacked on.

Where that earns its keep is in multi-stage sampling. A popular pattern for two-pass workflows - one sampler handling the high-noise steps, a second handling the low-noise finish - involves splitting a schedule into pieces and gluing them back together at the seam. Pad is the "glue" half of that: after you've cut a schedule down with Sigmas Truncate, you often need to re-attach a terminal value (frequently zero, marking "fully denoised") so the piece behaves correctly wherever it heads next. Sigmas Unpad is the companion node that removes it again later in the graph if you don't want it to stick around.

The inputs and outputs that matter

Two required inputs: sigmas, the schedule to pad, and value (a FLOAT, default 0, wide range) - what gets appended. There's one output, SIGMAS, the original schedule plus your new final entry, ready to wire onward.

The default of 0 lines up with the most common reason to reach for this node: a lot of downstream logic, this pack's included, treats a trailing zero as "fully denoised, nothing left to do." If you've cut a schedule down with Sigmas Truncate and whatever comes next expects to see that closing zero explicitly rather than just stopping wherever the truncated tail happened to land, Pad is how you put it back.

Installing it

ComfyUI Manager: search RES4LYF, install, restart. Manual path:

cd ComfyUI/custom_nodes
git clone https://github.com/ClownsharkBatwing/RES4LYF/
cd RES4LYF
pip install -r requirements.txt

On a portable ComfyUI build, use the embedded python_embedded\Scripts\pip.exe instead of plain pip. Restart ComfyUI and hard-refresh the tab afterward.

Where people get tripped up

The main confusion is step count drift. Padding adds an entry, which means whatever's downstream - a sampler expecting a specific number of steps, or another Sigmas- node doing index-based math - now sees one more than before. If a graph worked before you inserted Pad and breaks afterward, that extra entry is the first thing to check, and Sigmas Unpad is the fix if you added it only as scaffolding for something upstream (like Sigmas Math3) rather than something you actually want a sampler to see.