The Higher Resolution, the More Quality Downstream Tasks.

It is wonderful to utilize this tool to parse all the figures from a specific PDFs. Thank you for your great work. These days, I apply this to extract them from ACL papers related to ACL60/60 dataset.

However, the following problems could be found:

**❓Problem**
- Low Resolution
- Can't customize resolution (e.g. DPI)
- Rarely miss some figures even though 'Figure' caption is found

The issue names no files or tests. Start by locating the figure-extraction entry point and how output image resolution is determined; then clarify whether configurable DPI and missed figures are separate changes. Done should include a defined resolution setting and reproducible checks for the reported extraction cases, but the acceptance criteria need maintainer clarification.

Escrita pelo modelo de indexação a partir do texto da issue.