Add image differential annotation mode #99
Description
Add capability to annotate frame-to-frame differences in video/movie stimuli. This would take two frames (or a video + timestamps) and generate HED annotations describing what changed between them.
Requires research on:
- Visual change detection
- VLM prompting for difference description
- Relevant HED temporal tags (Onset, Offset)
See .context/image-differential-research.md for research findings.