Cloud Random Noise

When you build a custom sampler chain - model, guider, scheduler, sampler - one more input is almost always hiding in the corner: the noise. Somewhere you have to say what random noise tensor the sampler starts from and what seed produced it. In the cloud that's this node's job. Cloud Random Noise contributes a RandomNoise entry to the workflow JSON and returns a CLOUD_NOISE handle you feed into the sampler, exactly where a local RandomNoise node would sit.

It's the smallest node in the pack and it knows it: one input, one output, no opinions. But it's also the reproducibility knob. Same seed, same chain, same cloud job → same output (assuming the cloud worker behaves deterministically, which it mostly does). Different seed → different result. If you're iterating on a prompt, this is where you hold the seed fixed so you can actually compare changes.

Inputs

• noise_seed - INT, default 0, range up to 2^64. Set this and you control the starting noise for the run.

Output: noise (CLOUD_NOISE), wired into the noise 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, paste an API key from https://platform.comfy.org/profile/api-keys, restart. Manager-installable as "ComfyUI-CloudAPI-worker"; deps: requests, Pillow, safetensors.

Gotchas

• CLOUD_NOISE is a distinct handle type. It only plugs into cloud sampler nodes that expect it - don't try to feed it into a local RandomNoise input or vice versa. The pack's "sampling primitives" (this, Cloud Manual Sigmas, Cloud CFG Guider, Cloud Sampler Custom Advanced) are designed to be used as a matched set.
• Changing the seed without changing anything else is the cleanest way to explore a prompt space; zero is the default, not a special value.
• Proof-of-concept pack, so don't expect cloud-side guarantees of bit-exact reproducibility across retries - the seed gives you likely repeatability, not a contract.