Skip to content

(Cross-Platform) An open source approach to locally record and enable searching everything you view on any computer.

Notifications You must be signed in to change notification settings

jasonjmcghee/xrem

Repository files navigation

xrem

Cross-platform in-progress implementation of rem

Way too early to use- seriously, it's not useful yet.

But I want to help develop it!

Oh! That'd be awesome. Just run:

npm run tauri dev -- --release

Does a lot worse in debug mode in terms of performance.

"implemented" (read: likely terrible)

Currently implements, in a parallel / non-blocking way:

  • click the status icon and choose to start / stop recording
  • screenshot capture every 2 seconds
  • OCR at capture time
  • calculate text embedding (rem doesn't have this yet lol, and this is commented out anyway for now, but it works)
  • stream to mp4 without writing pngs to disk
  • basic tray icon + menu
  • efficient timeline seeking of a recorded data (with front-end)
  • view and "search" history as thumbnails: i put it in quotes because search is not working well yet
  • navigate to timeline frame by clicking search result

NOTE:

  • NO CACHING YET (this is vital for fast seeking between video files, currently big delay when swapping chunks)

(more recent screenshot of tray - still using default tauri logo)

image

Recent "search" (recent items) functionality

Screen.Recording.2024-01-19.at.5.13.04.PM.mov

More recent timeline seeking demo

Screen.Recording.2024-01-14.at.5.29.07.PM.mov

First Demo of basic poc:

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.

FAQ

  • 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

not implemented at all

  • Live OCR
    • overlay transparent text on image so as to be selectable (potentially works on Mac due to webview, but won't work elsewhere)
  • better search UI (matched text - not working well, date, application?)
  • natural language search
  • settings
  • window-specific OCR / filtering

About

(Cross-Platform) An open source approach to locally record and enable searching everything you view on any computer.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published