Hello,
I would like to report an issue in QuMagie regarding how videos are handled, and also propose some improvements.
Observed Issue
Unlike photos, which benefit from generated and cached thumbnails, videos do not seem to have any kind of pre-generated preview.
As a result:
Video loading in QuMagie is very slow, even for short clips.
The preview always takes several seconds to appear, which significantly affects the user experience.
Additionally, the AI recognition features (people, objects, scenes) appear to completely ignore videos, even though videos could contain relevant data for organizing content.
Suggested Improvements
I would like to suggest adding video preview generation settings similar to those already available for photos.
1. Multi-image thumbnail generation
Extract X frames evenly distributed across the duration of each video (e.g., 4 frames).
Combine these frames into a lightweight animated GIF to serve as the preview.
Provide quality settings (resolution, compression, number of frames), just like for photo thumbnails.
2. Time-based frame extraction
Extract one frame every X seconds of video.
Automatically generate a GIF or short animation from these frames, with a configurable maximum number of images per video.
AI Improvements
It would also be very helpful if videos were included in the AI analysis pipeline (people, objects, scenes), at least by analyzing some key frames, in order to improve content organization and classification



