CombineParamData

Most nodes in the mittimiLoadPreset2 pack do something visible: load a model, save an image. CombineParamData (class CombineParamDataMittimi) does one thing and one thing only - it merges two parameters_data blobs into one so a single image can carry two recipes in its metadata.

What it's for

When LoadSetParameters runs, it emits a parameters_data structure - a list of dicts holding the prompts, sampler, CFG, seed, model and size of that pass. Its sibling SaveImageWithParamText writes that structure into the image as an A1111-style parameter block. But what if your output is the result of two passes with different settings - an img2img source and result, a high-res fix, a refiner stage, a ComfyUI and A1111 combo you want to document? A single SaveImage only has one parameters_data input. That's the gap this node fills: feed it two PDATA structures and it hands back one combined PDATA, which SaveImageWithParamText then writes out with both recipes separated by a blank line.

Inputs and output

• param1 - required PDATA, the first recipe.
• param2 - optional PDATA, the second. Leave it unconnected and the node passes param1 through untouched.

The output is a single parameters_data that you wire straight into SaveImageWithParamText's parameters_data port. It's genuinely that simple - under the hood the node just concatenates the two lists, so if you chain several of these you can stack up any number of parameter sets before the save node.

Should you bother?

Only if you actually want multiple parameter blocks in the metadata. If you're saving plain single-pass outputs, you don't need this node at all - it earns its place specifically in the multi-stage or img2img workflows where "what settings made this?" has more than one answer. It's also handy as a passthrough: because param2 is optional, a lot of setups just run param1 → CombineParamData → SaveImage as a no-op connection that keeps the data path obvious. Either way, it's three inputs, one output, no hidden behavior.

Install

Part of the same pack, so the usual route:

cd ComfyUI/custom_nodes
git clone https://github.com/mittimi/ComfyUI_mittimiLoadPreset2

or install via ComfyUI Manager by searching "ComfyUI_mittimiLoadPreset2", then restart ComfyUI. The pack's only dependency is toml - no models, nothing heavy. If the source LoadSetParameters node is misbehaving because you also run RES4LYF, apply the scheduler-registration patch documented in the pack README (from ClownsharkBatwing/RES4LYF#161) before you chase problems in the data path - combine and save both just pass through whatever PDATA they receive.