Fast-RH makes remote RunningHub workflows feel more like local ComfyUI workflows. It provides convenient controls for LoRA models, KSampler parameters, image size, random seeds, and RunningHub configuration.

Built as a companion to the official ComfyUI_RH_APICall nodes — you can keep using the official workflow execution nodes.

简体中文 | English

Installation

Place this repository in ComfyUI/custom_nodes/ComfyUI-Fast-RH, or install it through ComfyUI Manager using its Git URL.

Fast-RH Settings

The official RH Settings node saves api_key as a node parameter. When you share an image containing workflow metadata, that key may be included in the image.

Replace RH Settings with Fast-RH Settings. The node has base_url, workflowId_webappId, and config_path inputs. Its apiConfig (STRUCT) output connects to the official RunningHub execution and upload nodes without changing the rest of the workflow.

Setup

Fast-RH Settings node

Copy config.example.json to config.json in the ComfyUI-Fast-RH plugin directory, then enter your RunningHub API key:
```
{
  "api_key": "enter_your_api_key_here"
}
```
In Fast-RH Settings, set base_url to the RunningHub site associated with that API key. Enter the remote workflow or Web App ID in workflowId_webappId.
config_path defaults to config.json in the plugin directory. For a file elsewhere, enter an absolute path or a path relative to the directory containing node.py. Connect apiConfig to the execution or upload node that previously received the official RH Settings output.

ComfyUI reads the API key from the file when it runs the workflow; the key is not saved in the workflow. After editing config.json, queue the workflow again to use the new key.

Note: Replace old RH Settings nodes before sharing workflows. Images and workflow files created earlier may still contain the original API key; replacing the node does not remove it from those files.

Fast-RH LoRA Stack

Create a Load LoRA placeholder in the remote RunningHub workflow first. You can select any model and set strength_model to 0. Then you can switch the remote model from ComfyUI.

Load LoRA placeholder in the remote workflow

How to use

Add Fast-RH LoRA Stack in ComfyUI.

Click Click to choose a LoRA model… in the node to open the model picker. Under Login Settings, choose the RunningHub site where you signed in, paste Rh-Accesstoken from your browser's developer tools under Cookies, and click Save Login. A full Cookie string or Bearer token also works.
Choose a model category, search for a model, and select its version. Click the version's cover to fill its model filename into the current LoRA entry.
In that entry, enter the ID of the corresponding LoRA loader node in the remote workflow. Set model strength and enable the entry. If the remote loader has a strength_clip input, turn on Send CLIP strength and set its value. Leave that switch off for loaders without this input. You can add up to 16 entries; each enabled entry needs a corresponding remote node ID.
Connect the node's nodeInfoList (ARRAY) output to the official RunningHub RH Execute Workflow node's nodeInfoList input. To combine other parameters, connect the preceding parameter node's ARRAY output to previousNodeInfoList first.

If the model list is stale, click Refresh List to update the current page or Clear List Cache to clear cached lists. Use Set Cover to choose a local cover image and Restore Remote Cover to undo it. If your login expires, sign in to RunningHub again and save a new token under Login Settings; Clear Login removes the locally saved token.

Fast-RH Random Seed

Fast-RH Random Seed node

How to use

Find the node ID whose seed you want to change in the remote workflow. Enter it in nodeId, then set seed.
In ComfyUI's control after generate for seed, choose randomize to use a new seed on each queued run, or fixed to keep the entered seed.
Connect nodeInfoList to the official RH Execute Workflow node's nodeInfoList input. To include other parameters, connect the preceding parameter node's output to previousNodeInfoList first.

Fast KSampler

Fast KSampler node

How to use

Find the KSampler node ID in the remote workflow and enter it in nodeId. Set seed, steps, cfg, sampler_name, scheduler, and denoise. To use a new seed on each run, set seed's control after generate to randomize.
Connect this node's nodeInfoList output to the official RH Execute Workflow node's nodeInfoList input. If you already use RH Node Info List or another parameter node, connect its ARRAY output to previousNodeInfoList first.

These settings override the remote KSampler parameters. This node does not run sampling locally.

Fast Empty Latent Image

Fast Empty Latent Image node

How to use

Find the Empty Latent Image node ID in the remote workflow and enter it in nodeId.
Choose Preset to select a width and height pair, or Custom to enter width and height separately. Use Swap Width / Height ↔ to exchange the current dimensions, then set batch_size. Width and height must be multiples of 8 from 16 to 16384; batch_size ranges from 1 to 4096.
Connect nodeInfoList to the official RH Execute Workflow node's nodeInfoList input. If you have other parameter nodes, connect their ARRAY output to previousNodeInfoList first.

Preset and Custom retain their own dimensions when you switch between them. This node only changes remote workflow parameters; it does not create a local LATENT.