Attribution research that shows its work
SourceTrace is an MVP for following publicly posted creative images across the web and turning scattered clues into an understandable evidence trail.
What SourceTrace does
SourceTrace accepts an uploaded image or supported public URL, builds safe search variants, collects candidate pages through replaceable providers, and ranks likely creators and sources using deterministic evidence.
It is designed first for artwork, fan art, gaming images, anime art, illustrations, and memes. This is media that frequently loses attribution as it is reposted.
What it does not do
SourceTrace is not a copyright registry, legal opinion, authorship authority, or facial-recognition service. A result is a research lead based on public evidence available at the time of the search.
Artwork is not training data
SourceTrace does not train models on submitted artwork, build generative-art datasets, or use uploads to reproduce an artist's style. The image is used only to perform the search requested by the user.
Live searching relies on bounded external API calls for visual search, visible-text extraction, and evidence review. Working files are metadata-stripped and retained with their search under the configured retention policy.
MVP status
The local fixture workflow remains available without external services. Live mode uses replaceable search and text-reading APIs, independent visual verification of returned candidate images, and bounded inspection of accessible public pages. Coverage is incomplete, inaccessible pages are not bypassed, and an empty search returns an inconclusive report.