Legion: Master (6 channels)

LegionMaster6 is the middle child of the Legion Master family: six inputs, six outputs, and the same legion_campaign handle as the 3- and 12-channel siblings. It exists purely so you don't have to stare at twelve mostly-empty ports when your worker only needs a handful of values.

You'd pick this over Master3 when the job legitimately wants more than three things shipped over. A worker that takes an image plus a mask plus a few numeric settings (strength, denoise, tile size) fits six channels comfortably without spilling into the sprawling 12-channel version. Like all the Master variants, it's the same node underneath - same config object, same sync/async logic, same serialization, just a different number of ports exposed.

Inputs and outputs

• legion_config (optional) - from Legion: Configuration. Either this or legion_campaign, not both.
• legion_campaign (optional) - reuse a campaign/worker instead of a fresh config.
• input_1 … input_6 (optional) - data to send to the worker; empty ports are skipped.
• Outputs: legion_campaign, plus output_1 … output_6 - results in sync mode, error strings in async mode (collect via Legion: Join).

The usual pack constraints apply: only images and primitives cross the wire, and the worker workflow must contain Legion: Importer at the start and Legion: Exporter at the end.

Installing

Manager search "LegionPower", or:

cd ComfyUI/custom_nodes
git clone https://github.com/Transhumai/ComfyUI-LegionPower.git
cd ComfyUI-LegionPower
pip install -r requirements.txt

Restart. Just requests as a dependency.

Troubleshooting

Same book as every Master: Worker failed to start in time (check the console for the launch command, ports, ComfyUI path), Output manifest not found (worker workflow missing a connected Exporter), and None outputs from unsupported data types. Test with dry_run: true if you want to verify the plumbing without spending time booting a worker.

Rule of thumb: 3 channels for simple jobs, 6 when you're passing a few settings alongside your images, 12 only if you genuinely move that much data. Most people will land on 3 or 6.