OpenOutpainter Serving Style Get

Style Define publishes styles to the canvas dropdown. This is the other half: when a user actually checks one of those styles and generates, the style's name rides along inside the request - and OpenOutpainter Serving Style Get is how your workflow finds out which one was picked and pulls out its prompt text.

It's a conditional node with a lookup baked in. Required inputs:

• oop_request - the request from the Serving node.
• style_name (STRING) - which style you're looking for. If you want several, use several Style Get nodes, one per name.
• empty_strings_on_false (BOOLEAN) - per the tooltip, "otherwise blocks execution on the prompt outputs": when true, a no-match returns empty strings; when false, the prompt outputs get an ExecutionBlocker and the branch stops.

Outputs: oop_request_if_true / oop_request_if_false (the request routed by whether the style was selected), boolean (the match result), and prompt / negative_prompt (the style's saved text).

How it works

The request carries a styles list - the names the canvas user checked. The node checks two things: that style_name is in that list, and that it exists in the styles you defined. The defined styles come from the Style Define chain, which the Serving node stashes on the request's extra data; that's why Style Get finds them without you wiring the style list directly into it. If both checks pass, you get the request on the true output plus the style's prompt and negative_prompt on their outputs, ready to merge into your CLIP text encoding. If not, you get the false output and - depending on that toggle - empty strings or blocked execution.

Practical pattern: chain a Style Get off each Input node in your generation branch, wire its prompt output into a CLIPTextEncode alongside the request's own prompt (say, concatenated), and combine the boolean with other conditionals if you want. The routing outputs (oop_request_if_true/false) matter if a checked style should change which branch runs - e.g., a "detail upscale" style that routes to a higher-step branch.

Install and the snags

Pack install, once: ComfyUI Manager (search "OpenOutpaint ComfyUI Interface") or git clone https://github.com/without-ordinary/openoutpaint_comfyui_interface into ComfyUI/custom_nodes, restart. Only opencv-python in requirements; nothing to download.

First snag: Style Get returns nothing useful if the style list never reached the Serving node - so verify your Style Define chain is actually wired into its optional oop_styles input, or the lookup table is empty no matter what the canvas shows. Second: remember this node only reports a selection; nothing merges the style's prompt into the generation unless you wire the prompt/negative_prompt outputs into your text encoding yourself. And the pack-wide rule stands - one serving workflow at a time.