⚙️ RH Param
Say 'set the seed to 42' to a cloud workflow — RH Param
- previous_params
- params
RH Param is how you tell a RunningHub workflow what to actually do. The cloud workflow has its own saved inputs - prompt, seed, steps, whatever - and by default RH Execute runs it exactly as saved. This node edits one field on one node of that workflow per instance, and you chain them together to override as many as you like.
It's a small, single-purpose node, and that's the appeal: one Param per knob, laid out on your canvas, so your "change the prompt, keep the seed" logic is visible as a graph instead of buried in a JSON blob.
The three inputs that matter
node_id- the node number inside the RunningHub workflow (like3or47). This is the bit people fumble. It is not the ComfyUI node's ID on your local canvas; it's the ID of the node in the cloud workflow you're calling. The RunningHub editor shows these numbers on the nodes; "Run with API" will surface them too.field_name- a dropdown of the common field types (text, image, video, mask, seed, steps, cfg, sampler_name, scheduler, denoise, width, height, batch_size, model, vae, lora, control_net, strength, scale…). It's a convenience menu, not a contract: the actual field name has to match what the cloud node expects. If your field isn't in the list, pickcustom.field_value- the value, a multiline string. Yes, everything is typed as a string - that's just how the RunningHub API takesnodeInfoList. "42" for a seed, "5" for steps.
Two optional inputs round it out: previous_params to chain (see below) and custom_field_name, used only when field_name is set to custom so you can type an arbitrary field name.
Chaining and mechanism
Internally each Param builds a small dict - {nodeId, fieldName, fieldValue} - and appends it to a list. The params output is that list, which eventually feeds the params input of RH Execute. To set five things, you don't need five wires into Execute: wire Param A's params output into Param B's previous_params, and so on down the chain. The last one carries the whole list. Each node also prints what it set to the console, so you can audit the chain as it builds.
When you'd use it
Any time you want a cloud run that isn't the saved default. The classic first workflow is: RH Config → RH Param (set text to your prompt) → RH Execute → Preview Image. Add a seed Param and a steps Param and you have a t2i rig. The upload nodes (RH Upload Image, RH Upload Video, etc.) build the same nodeId/fieldName/fieldValue entries themselves, and their params output chains into the same list - so an image you upload becomes just another entry in the chain.
Gotchas
The dropdown lies a little: it's a picklist of common field names, and the cloud workflow may name its inputs anything. When in doubt, check the actual field name on the RunningHub side and use custom. Values are strings, so leading/trailing whitespace is real - a stray space in a seed is a subtle way to get "different" results. And note RH Param only sets inputs; it can't touch the workflow's internal graph.
Install is the pack-wide two-step: ComfyUI Manager (search ComfyUI_RH_API) or git clone + pip install -r requirements.txt, then restart.
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| node_id | STRING | Node ID in the workflow (e.g., '3', '47') | |
| field_name | COMBO | text | Field name to modify - select from common types or use 'custom' |
| field_value | STRING | Value to set for the field | |
| previous_paramsopt | RH_PARAMS | Connect previous RH_Param node to chain parameters | |
| custom_field_nameopt | STRING | Custom field name (only used when field_name is 'custom') |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| params | RH_PARAMS | — |