Split from #1651.
Allow forcing output images to a lower resolution/DPI when the input was scanned at an excessively high DPI.
Important nuance on why (and why this is not primarily a file-size feature): aggressive lossy compression (JPEG/JPEG2000 quality) already behaves like a "smart" downsampling — it allocates bits where detail exists — and generally gives better quality for the same bit budget than uniformly reducing DPI. Most users should set their optimization to a more aggressive setting instead of downsampling, which is why it wasn't implemented yet.
The real benefit of capping DPI is reducing memory and processing cost when the input DPI is excessive (very large pixel dimensions), plus producing sensibly-sized images for downstream consumers.
Current state: only --oversample (an upsample floor for the OCR input) and --ghostscript-jpeg-maxdpi (PDF/A path only) exist; there is no general "downsample output to N DPI."
Split from #1651.
Allow forcing output images to a lower resolution/DPI when the input was scanned at an excessively high DPI.
Important nuance on why (and why this is not primarily a file-size feature): aggressive lossy compression (JPEG/JPEG2000 quality) already behaves like a "smart" downsampling — it allocates bits where detail exists — and generally gives better quality for the same bit budget than uniformly reducing DPI. Most users should set their optimization to a more aggressive setting instead of downsampling, which is why it wasn't implemented yet.
The real benefit of capping DPI is reducing memory and processing cost when the input DPI is excessive (very large pixel dimensions), plus producing sensibly-sized images for downstream consumers.
Current state: only
--oversample(an upsample floor for the OCR input) and--ghostscript-jpeg-maxdpi(PDF/A path only) exist; there is no general "downsample output to N DPI."