FAQ and Troubleshooting

FAQ

How is this different from tools that export each page as one image?

The output goal is different.
Those tools mainly optimize for looking similar. PDF2PPT tries to preserve the original appearance while rebuilding text and some image regions into objects that remain editable in PowerPoint.

Does it guarantee a fully editable PPT?

No.
The final level of editability depends on scan quality, page complexity, OCR quality, and image-region detection. In practice, the safer strategy is to preserve visual fidelity first, then improve editability where possible.

Do I have to configure a remote OCR API key?

Not necessarily.
The project supports both local and remote OCR/parsing pipelines. But if you want remote OCR as the default path, you need to fill the corresponding variables from .env.example.

Is it better for local use or service deployment?

Both are supported.
For personal usage, local development or single-machine Docker is enough. For shared use across a team or other systems, the standard Docker deployment is the better fit.

Useful Troubleshooting Commands

docker compose ps
docker compose logs -f api
docker compose logs -f worker
docker compose logs -f web

Troubleshooting Directions

• OCR gets stuck: inspect worker logs, job status, and OCR debug artifacts
• OCR is correct but part of the output becomes an image: inspect image_regions and scanned-page reconstruction strategy
• First startup is slow: check Paddle model prewarm and cache download behavior
• Web works but API does not: check NEXT_PUBLIC_API_URL, INTERNAL_API_ORIGIN, and health checks