[PDFBOX-3763] Optimize SampledImageReader.getStencilImage()

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.5
• Fix Version/s: 2.0.6, 3.0.0 PDFBox
• Component/s: Rendering
• Labels: optimization
• Assignee: Tilman Hausherr
• Reporter: Tilman Hausherr
• Created: 19/Apr/17 16:42
• Updated: 15/May/17 20:28
• Resolved: 19/Apr/17 17:05

Description

I have a PDF file (which I can't share, it is p182 of the one from https://stackoverflow.com/questions/43248021/pdfbox-2-unusual-memory-consumption ) that needs over 3GB for rendering due to two stencil images that are 18694 x 13214. PDFBox converts stencil images to RGB images, where 1 pixel needs 4 bytes, so such an image uses 1GB.
After many dead ends over the easter weekend, I realized today that the BufferedImage isn't needed at all - we can process the bits directly.

Memory footprint is down from 3.6GB to 2.2GB and it is faster too.

Issue Links

• duplicates: PDFBOX-2258 Speed up SampledImageReader#getStencilImage()