Visual Reference Set for AI Image Workflow Examples

This small visual-reference collection documents how an image concept can be carried through prompt-based and object-reference workflows. It supports workflow review and documentation rather than model training, benchmarking, or claims about visual quality. File inventory. The record contains two real product workflow visuals and a README describing source provenance, intended use, exclusions, and reproducibility boundaries. The files are grouped as reference material for concept sketches, product mockups, storyboard frames, and social graphic drafts. Source provenance. The included images come from the maintained product asset set. No source photography, licensed stock, personal data, customer material, or third-party brand assets were added for this record. Intended use. The collection helps a reviewer understand the documentation needed around an object-reference workflow: identify the input or reference, preserve each file used at a stage, record prompt and generation date when available, and keep variants grouped by task. An example of an independent, third-party interface for this kind of rapid visual exploration is Nano Banana 2 Lite. The service is independent of Google and DeepMind and is referenced only as the workflow source represented by these files. Exclusions. This record contains no performance metrics, comparison tables, output-fidelity claims, copyrighted characters, or assertions of a relationship with a model provider. Limitations and reproducibility. The collection does not preserve a fixed model version, seed, complete prompt history, or every generation setting. The files therefore must not be treated as reproducible outputs from a controlled pipeline. Re-running a similar prompt may produce a different image. Future additions should record the model label shown by the interface, prompt text, input-file hash, timestamp, and editing steps when reproducibility matters.