SCG Image Result

This is the smallest honest node in the scg_json_prompt_agent pack, and it exists to answer one question: when you hit Run Workflow inside the SCG Ideogram4 Prompt Agent builder, which image should show up in the output tab?

By default the builder grabs the last Preview/Save image of the run. That's fine for a simple text-to-image graph. But the moment your workflow gets interesting - a preview at base resolution, then an upscaler, then a color grade - the "last preview" isn't the image you care about. That's where SCG Image Result comes in. It's a marker node: you wire your actual final image through it, and the builder's Workflow output tab shows that exact render, not whatever happened to be previewed last.

How it works

Under the hood it's a subclass of ComfyUI's own PreviewImage, re-categorized into SCG/Ideogram. So it inherits the exact same behavior - it saves a temporary file and reports it in the node's ui output - and the builder's parent extension (which talks to ComfyUI over the websocket) picks it up from the executed event by node type. The plumbing matters here: the builder iframe can't talk to ComfyUI directly, so everything it displays comes through that extension, and this node is simply how you tag which result it should pay attention to.

Because it behaves like Preview Image, the output is temporary. Wire it into your graph anywhere after a VAE decode - or after your upscaler, which is the usual reason to bother - and it passes the image through untouched.

The inputs and outputs

There's really only one of each, so this won't take long:

• images (IMAGE) - required input; your final render
• images (IMAGE) - output, the same image passed straight through

That's the whole schema. It's an output node (is_output_node), meaning nothing downstream of it runs - it's a terminal, exactly like Preview Image. If you want to actually keep the file, pair it with a Save Image node rather than relying on this.

Installing it

It ships inside the same pack as the builder, so install is identical:

cd ComfyUI/custom_nodes
git clone https://github.com/SanDiegoDude/scg_json_prompt_agent.git

Restart ComfyUI afterward (the pack registers its server routes at startup), and you'll find SCG Image Result under SCG/Ideogram. No extra dependencies, no model files, no API keys - this node touches no network at all.

Troubleshooting

• The output tab shows the wrong image. This node is the fix. Put it at the end of the branch you consider "final" - typically after your upscaler - and the builder will display that render instead of the last preview.
• Nothing changed after adding it. Make sure the node is actually in the graph that gets queued, and that the workflow run you triggered passes through it. An orphaned node does nothing.
• It's not a Save Image replacement. The output is temporary, by design. If you're hunting through your ComfyUI temp folder later wondering where the image went, this is why - add a Save Image if you need persistence.

For a utility node, it earns its place: one wire, and the thing you're iterating on shows the thing you actually made.