What this covers
- Greyscale conversion saves far more than compression does
- Delete blank pages first - it is free and often large
- Compressing twice degrades quality for almost no further saving
- Rescanning correctly beats processing a bad scan repeatedly
A folder of phone photographs of A4 pages is 2 to 5 megabytes each. Twelve pages is comfortably over every submission limit anyone uses. This is the most common practical problem in handwritten submission and it has a clear order of operations.
What each step saves
Measured on twelve photographed A4 pages, starting at 34.8 MB.
| Step | Result | Cumulative saving |
|---|---|---|
| Original photographs | 34.8 MB | - |
| Delete two blank pages | 29.0 MB | 17% |
| Greyscale with contrast correction | 1.8 MB | 95% |
| Resolution cap at 1800 px | 1.2 MB | 97% |
| Black and white instead of greyscale | 0.4 MB | 99% |
| Compression applied to the greyscale version | 1.6 MB | Barely anything further |
The last row is the important one: compressing an already-cleaned scan saves almost nothing, because the clean-up already removed what compression acts on.
Why greyscale does most of the work
A colour photograph of a page stores three channels of information about a surface that is essentially two-tone. Most of that data describes the exact shade of white of the paper under indoor lighting, which is not information anybody wants.
Greyscale conversion with contrast correction throws away the colour and normalises the background towards white, which both shrinks the file dramatically and makes the page more readable. It is the rare change that improves two things at once.
Do not compress twice
Each compression pass compounds the quality loss, and the second pass saves very little because the first already removed the redundancy. If a compressed file is still too large, go back to the original photographs and process them differently rather than compressing again.
The order to try things
Stop as soon as you are under the limit.
- Delete pages you do not need
Blanks, duplicates, a page you photographed twice. Free, and often ten to twenty per cent.
- Convert with greyscale clean-up
This is the big one. Typically takes a scan to a tenth of its size.
- Lower the resolution cap
2200 pixels on the longest edge is about 200 DPI on A4, which is ample. Dropping to 1800 is still readable and saves a third again.
- Try black and white, then check
Another factor of four, at the cost of faint pencil. Read the faintest page before accepting it.
- Only then, compress
And expect a small saving, because the clean-up has already removed most of what compression works on.
The tools, in that order
Why compression underperforms here
Compress PDF recompresses embedded images at a lower resolution and quality. On a raw colour photograph that is effective. On a page that has already been converted to greyscale at a capped resolution, most of the redundancy is gone, and the remaining saving comes at visible cost to the writing.
This is why the order matters. Clean-up first, compression last, and frequently not at all.
The better answer: rescan
If a document is still too large after all of this, the capture was the problem. A photograph taken in dim light with the camera far from the page has a lot of noise in it, and noise is expensive to store and impossible to compress well.
Reshooting twelve pages takes six minutes. Processing a bad scan through four successive tools takes longer and produces something worse.
Two routes to a small file
Both get under 2 MB. One is readable.
Clean up properly
- Greyscale with contrast correction
- Resolution capped at about 200 DPI
- No compression pass at all
- Around 180 KB per page
- More readable than the original photograph
Compress repeatedly
- Colour photograph compressed hard
- Then compressed again to get under the limit
- Visible artefacts around every stroke
- Around 180 KB per page
- Noticeably worse than the original
Getting under a limit without wrecking the document
- Delete anything you do not need first
- Greyscale clean-up before anything else
- Cap the resolution rather than compressing
- Try black and white only after checking the faintest page
- Compress at most once, and only if still needed
- If it is still too large, reshoot rather than reprocess
- Keep the original photographs until the submission is accepted
Questions and answers
Why did compression barely help my scan?
Because the scan had already been cleaned up, and clean-up removes most of what compression acts on. Compression is effective on a raw colour photograph and nearly pointless on a processed greyscale page.
What resolution do I actually need for handwriting?
About 200 DPI, which is roughly 2200 pixels on the longest edge of an A4 page. Higher adds size without adding anything a marker can see.
Is it better to compress or to scan again?
Scan again, if the original capture was poor. Six minutes of reshooting beats four processing passes on a noisy photograph, and the result is better rather than worse.
My portal limit is 10 MB and my file is 12. What is the quickest fix?
Check for blank pages first, then lower the resolution cap. Both are lossless in any way that matters for handwriting, and either alone usually clears 2 MB.
