Video Prompt Controlled-Pair Study Pack v1.0

24 original AI-assisted synthetic prompt examples arranged into 12 controlled pairs for planning text-to-video and image-to-video comparisons. This is an experiment-design resource, not an evaluated benchmark. No video generations or empirical ratings were produced. Six pairs use text-to-video and six use image-to-video. Within each pair, the base scene stays fixed while one written instruction changes: camera movement, motion, lighting, focus, or framing. All records are marked not_run. The ZIP contains prompts.csv, prompts.jsonl, observations-template.csv, README.md, and LICENSE.txt. The observations template is empty. Documentation describes input controls, a proposed manual rubric, reproducibility limitations, and licensing. No reference images, generated videos, personal data, or third-party prompt quotations are included. These examples were newly authored with AI assistance; they are not an export of the Seadanse Prompt Library. CC0 1.0 applies to the pack’s original text and data only, to the extent rights exist. It does not grant rights to third-party services, reference images, or future outputs. Created for the Seadanse project. The resource is model-agnostic and does not require a Seadanse account. Companion distribution: Public source files; Kaggle dataset.