Loading...
Loading...
Numonic reads and indexes the metadata inside your AI-generated assets. Prompts, seeds, samplers, and model versions, all preserved and all searchable, so the creative context behind a file is never lost.
Each field is captured on ingest and kept with the asset. Nothing here is retyped by hand, and nothing here is lost when the file moves.
Plus workflow graphs, node configurations, and embedding references.
Find assets by any metadata field. Search for a phrase inside a prompt, or pull every image that used one specific LoRA.
Each asset stores exactly what you need to recreate it. Same seed, same parameters, same result.
The record becomes the basis for organization. Group by model, filter by style, sort by parameter.
Know exactly which version of which model created each asset, so you never have to guess which checkpoint it was.
Full prompt and parameter extraction runs on images and audio, plus technical metadata for video (codec, bitrate, and container fields). Video prompt and parameter parity is on the roadmap.
Full workflow graphs, node parameters, prompts, and model configurations, read from the PNG metadata chunks. Video outputs contribute technical metadata from the MP4 atoms, plus codec and bitrate.
Full supportPrompt text, job IDs, parent references for variations, and generation parameters, read from the exported data.
Full supportEvery prompt, every parameter, every seed, captured on ingest and kept with the asset.