Recording generated-asset provenance, originals, and selection decisions
Generating assets with AI also produces candidates that never get used. Saving only the images makes it harder to reconstruct the references used, whether a person approved a candidate, and which processed file the game actually loads. A stable candidate ID can connect the original, generation parameters, reviews, and code references while keeping those checks separate. This article examines VOLT…
The article discusses the importance of recording provenance, originals, and selection decisions in AI-generated assets for video games. It highlights the challenges of tracking the reference images used, approvals, and processing details when generating assets with AI. The VOLT NOMAD team created a provenance document and a candidate-level manifest to address these issues.
The manifest groups candidates into generation batches and stores information such as candidate identity, original file prompt, model, dimensions, seed, generation parameters, reference image, generation status, timestamp, usage, errors, human review decisions, ratings, and review timestamps. This structure enables the separation of successful generation from the selection decision and allows for a traceable relationship between the original asset and the processed file used in the game.
The article provides an example of tracking a specific candidate, final-fallen-machine-seraph-v9-c, through its original image, generation parameters, and the final cutout used in the game. While the manifest confirms the selection decision, the file loaded by the game differs from the original asset, demonstrating the need for accurate provenance records.
Brief written by urgent.news from Dev.to's own syndicated text. Machine-written — may contain errors; check the original before relying on it.
