Turbo Workflow Output

Let's get the confusion out of the way first: "Turbo Workflow Output" sounds like the node that saves your Krea 2 Turbo renders. It doesn't. It produces no image and no latent. It's a verification node that audits your Turbo workflow and publishes a receipt - a model-free artifact bundle with fingerprints and a sampler receipt that explicitly says nothing was executed.

So why would you reach for it? For reproducibility and provenance. If you're sharing a Krea 2 Turbo workflow, or tracking experiments and want to prove "this run was on the strict official 8-step schedule, on this exact host, and I am not claiming I ran a sampler," this is that proof. For daily generation it's overkill. For people who obsess over audit trails, it's the entire point.

How it works

Under the hood it's a strict verifier. You feed it three things: the sigmas from a Sigmax Krea 2 scheduler, its schedule_info JSON, and a matching schedule_report. The node rebuilds the authoritative Krea 2 Turbo schedule from the dimensions embedded in schedule_info, compares it against what you actually connected - fingerprints and all - and refuses to run if anything differs. Specifically, it demands the official Turbo profile (krea2.turbo.official), strict_official on, and the complete 8-step slice (start 0, end 8). Change the step count and it errors out rather than quietly blessing a modified schedule.

When everything matches, it publishes a portable execution bundle whose receipt is stamped NOT_EXECUTED with zero effective transitions. That's honest, and kind of refreshing: it never runs the sampler, never touches a model, and it says so right there in the receipt.

The name is a lie in one more direction - it's Krea 2 Turbo-specific. Don't feed it a Wan or Z-Image schedule; it will reject them.

Wiring it up

All three inputs are required:

• sigmas (SIGMAS) - from Sigmax.Krea2SigmaScheduler.
• schedule_info (STRING) - the second output of that same scheduler.
• schedule_report (STRING) - the report from Sigmax.ScheduleInspector, fed with those first two.

So the chain is Krea2SigmaScheduler → (sigmas + schedule_info) → ScheduleInspector → schedule_report, then all three into this node. It's an output node, so it has no outputs to wire onward - its sigmax_execution_bundle JSON lands in the UI output / prompt history, where you can inspect or export it.

Install

Same pack, same steps: search ComfyUI-Sigmax in ComfyUI Manager and restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/rookiestar28/ComfyUI-Sigmax comfyui-sigmax

Restart ComfyUI. No extra dependencies and no model downloads - this node is deliberately model-free.

Where people get burned

The failure mode is a mismatch error when the inputs disagree: if schedule_report doesn't match the connected sigmas/schedule_info, or the schedule isn't the strict official 8-step Turbo recipe, the node raises rather than fudging it. That's not a bug - it's the point. If you built your workflow with the RAW recipe or bumped steps to 10, this node will (correctly) refuse to bless it.