Cross-platform in-progress implementation of rem
Way too early to use- seriously, it's not useful yet.
I don't care
Fine. Just run:
cargo run --release
Does a lot worse in debug mode in terms of performance.
Currently implements, in a parallel / non-blocking way:
- screenshot capture every 2 seconds
- OCR at capture time
- calculate text embedding (rem doesn't have this yet lol)
- stream to mp4 without writing pngs to disk
Taking screenshots + ocr transcript printed to terminal, screenshots streamed to videos and saved, then manually showing this by scrubbing.
out.mp4
It can keep up on my M1 Air, haven't tested elsewhere yet...
BUT, this is using cross-platform rust libraries for the functionality we need.
- when will it be ready?
- idk, but with your help it might go faster
- why is the current ocr solution screaming about ARNs?
- idk, if it was AWS's textract it might make more sense
- not having live text analysis (MacOS) seems like selecting from past screenshots will be much harder to build
- yeah
- DB layer -> talking to duckdb
- frame extraction by index from mp4 videos
- timeline UI / full screen ui leverage frame extraction
- Live OCR
- overlay transparent text on image so as to be selectable
- winit + wry? (html / js / css)
- winit + iced? (rust)
- Live OCR
- search UI (display thumbnails, matched text, date, application?)
- settings
- tray icon + menu
- window-specific OCR / filtering