Closed DmacMcgreg closed 4 days ago
The way I see the app having value is that if you were to focus on having a timeline feature like rewind, or ScreenMemory, you could provide key points where you were using apps, and highlight them on the timeline. The same with searches. If you search gmail, then you highlight all the points on the timeline, where you could easily scroll back to those main points on the timeline to see what gmail was doing. Right now, the focus on OCR text is providing too much noise. Also since it splits the video files arbitrarily into small chunks, its hard to actually see what the purpose of the search results are since they are split up among many sections that have a lot of duplicate data.
@DmacMcgreg thanks for the feedback!
I created a bounty here https://github.com/mediar-ai/screenpipe/issues/609
would love any more context on it, how to make it great
regarding OCR we're going to release today a new alternative data source which extracts text natively from the UI and will be unique and highly structured
This is great you're opening a bounty for this one. Some ideas:
describe the bug When I open the app, its difficult to find any use, since the summaries it gives me are just walls of gibberish. (I've blurred some Personal information)
to reproduce
expected behavior The text should be formatted, correct and useful. Right now, I'm having a hard time finding any value in this setup.
system info
additional context