bquenin / interpreter Goto Github PK

Hi! I just want to first let you know that this is my favorite OCR! Unfortunately, it has stopped working for me with PPSSPP as well as OBS Studio. The last time I opened the app was October 14th, and it worked fine then. Maybe it has to do with the most recent update? When I try to open Interpreter, it seems like it's looking for the window, but then it will shut down. I'm looking to hearing from you soon for a resolution hopefully!

Whenever I'm in game on Heaven Burns Red and open interpreter to translate the game, it will only translate what it captures first when opened. If I advance one line forward it will no longer translate the new line of text or subsequent lines after that. Curiously, if you close interpreter, advance one line of text, and reopen interpreter it will still only have what it translated the first time, but if you close the game and the interpreter it will translate the new line of text. Also, in the subtitles for the translated text, it captures the text that comes from interpreter which i don't think is supposed to happen. Lastly, this the only game that doing this. All the other games I've tried with interpreter have worked and Heaven Burns Red is the only one giving me problems.

Using this image as a test, the program fails to recognise any of the text in the browser.

I checked what the screenshot comes out as, and it's a completely black image. The dimensions seem correct though.

OS Build is 22622.290 (22H2).

Edit: In mpv, the screenshots actually show something but they don't capture the window correctly. It instead shows my browser and a part of Windows Terminal that were beneath the window. It's extra weird since the tab that's showing was closed when the screenshot was taken, and I had also closed Terminal as well. At least the text recognition is working.

Edit 2: The screenshot appears to take what was underneath the window the moment the target program was opened, but it doesn't update what's displayed on the windows underneath, which explains my situation above. But sometimes it does take a correct screenshot, but it still doesn't show the updated window. Idk, I have no experience with Win32 APIs so my fiddling around hasn't really accomplished much.

Just curious how much is each request (or sets of request) end up costing for this? Are results cached? Looking at the viability in the long term

i really like the idea of this program, seems promising so far. however unfortunately it's pretty unusable in its current state. i tried to translate this dialogue from a game but the only a portion of the translated text is visible. it gets cut off because it's putting it all in one line like this:

for context, here's what the game was displaying:

i think more customization for the output text would help a lot to make this usable. line breaks are definitely something i think this program needs to avoid issues like this, as well as changing the size of the text to allow more text to fit on the screen. ideally the program could maybe overlay itself over the JP text and try to match the original text in size and position (similar to the way Google's mobile app does when you scan text) but i'm not sure how complex that would be to implement

https://github.com/tesseract-ocr/tesseract