Feat: transform pdf files into images to skip OCR

[Pascal-Code]

Hi,
i am using Qwen 3.8 27b locally for my airgapped librechat instance.
Unfortunately I can´t use any local ocr currently. Qwen is very good in analyzing images but it can´t handle scanned PDF files.

Would it make sense to convert a scanned pdf into images before it is send to the model?

This also makes sense if the PDF contains images or graphs to be analyzed. OCR does not work here properly.
Currently I have a workaround as a skill that does it with the code interpreter but it does not work reliable.

BR, Thanks

[Pascal-Code]

I see there was already a similar discussion to upload files directly to the model. However this does not work for all models. #4503