Upload Artifact Options

Upload Artifact Options is the settings-bundle node of the Embeddr pack. It doesn't upload anything itself - it builds an EMBEDDR_UPLOADARTIFACT_OPTS object that you wire into Upload Artifact or Upload Video to tell those nodes how to upload. Think of it as a context object, the same pattern that runs through rgthree's Context and the whole pipe-node family in ComfyUI: several values packed into one object that travels down a single wire, instead of five dangling inputs on every upload node.

You only need it when you want to control the details. Plain "just save it" uploads work fine without it - that's why the options input on the upload nodes is optional. The moment you care about tags, storage location, or whether the library should process the file, this is the node.

The five knobs

• storage_provider / storage_path - overrides for where the artifact is stored (the default storage comes from your Embeddr server). Leave both empty for the default; set them only if you run multiple storage backends or a custom layout. The upload node treats empty/__default__ as "use the default," so an empty field is safe, not an error.
• trigger_ingest - defaults to true. This is the interesting one. It controls whether the backend runs the ingest process after upload - in practice that means the embedding/compute step that makes the artifact searchable via the Find Similar nodes and enables automation. Turn it off and your upload goes in as a dumb file: stored, not searchable, no automation. For a batch of 500 reference images you don't care to search yet, disabling it is a legit speed play; for anything you'll want to find later, leave it on.
• tags - tags to attach to the uploaded artifact. The field is a string; the node splits/normalizes it into a clean list. Tags are what make the library navigable beyond collections, so it's worth a habit of tagging uploads.
• related_artifact_ids - IDs of related artifacts, attached as related/collection references on the upload. Not the same as parents - these are "things this is related to," used for grouping - but the upload node folds them into the same metadata bundle.

The single output, options, is the object that plugs into the options input of Upload Artifact and Upload Video.

How it works under the hood

Simple and honest: each input becomes a field on an EmbeddrUploadArtifactOptionsObject, and that object travels to the upload node, which reads the fields back when it builds the upload metadata. The upload node maps trigger_ingest onto both the trigger_automation and compute_embedding metadata flags (legacy compatibility), and turns related_artifact_ids into collection_ids - so "related" becomes "in these collections" on the backend. Worth knowing if you're debugging why something landed in an unexpected collection.

Install & notes

Standard pack install: embeddr serve running, pack via ComfyUI Manager or a release into custom_nodes, restart ComfyUI. No network calls of its own, no extra dependencies - it's a pure value node, so there's basically nothing to break. The one real caveat is that it's optional by design: if you wire an options node in and later remove it, the upload nodes fall back to defaults seamlessly, so experiment freely. Just remember an options object left connected to the wrong upload can carry tags or storage overrides you forgot about - the graph is doing what you told it, even the parts you stopped thinking about.