Cloud Manual Sigmas

Most of the time you let a scheduler pick your sigmas. Sometimes - debugging, replicating someone's exact result, or chasing a specific look - you want to type them yourself. Cloud Manual Sigmas is that escape hatch for the cloud: it contributes a ManualSigmas step to the workflow JSON and hands back a CLOUD_SIGMAS handle you feed into the sampler, exactly like a local ManualSigmas node but as part of a Comfy Cloud job.

It's a power-user node, and it knows it. The input is a single STRING of comma-separated sigma values - 1, 0.5 is the default - and the node does no validation magic. What you type is what the sampler walks down. If you don't know what a sigma sequence should look like, this node isn't where you learn; if you're porting a workflow that came with a hand-tuned schedule, this is the faithful way to reproduce it cloud-side.

Inputs

• sigmas - STRING, comma-separated noise levels, default "1, 0.5". High-to-low. The number of values effectively sets your step count.

Output: sigmas (CLOUD_SIGMAS), wired into the sigmas input of a Cloud Sampler Custom Advanced.

Installation

Standard pack install:

cd ComfyUI/custom_nodes
git clone https://github.com/Dobidop/ComfyUI-CloudAPI-worker

Copy config.json.example to config.json, add an API key from https://platform.comfy.org/profile/api-keys, restart. In ComfyUI Manager under "ComfyUI-CloudAPI-worker"; deps: requests, Pillow, safetensors.

Gotchas

• Garbage in, garbage out. A malformed or nonsensical sequence fails cloud-side (or worse, produces a mangled image without erroring). The defaults in the README examples are a safe starting point.
• This is the manual alternative to Cloud LTXV Scheduler - the LTX path is where most people land, since flow-matching models usually want their shift-based schedules rather than a hand-typed list.
• The pack is a proof of concept, and the sampling-primitive nodes (this, Cloud Random Noise, Cloud CFG Guider, Cloud Sampler Custom Advanced) are the "assemble your own sampler" subset. They're intended to be used as a set - mixing a CLOUD_SIGMAS from one pack with a cloud sampler from another won't typecheck.