# Scanned PDF Too Large? Rescan or Shrink It Right - Author: FastBankConvert (https://fastbankconvert.com) - Published: 2026-10-17 - Category: Speed, size and limits - Canonical: https://fastbankconvert.com/blog/scanned-pdf-too-large > Scanned PDF too large to upload? Resolution, color mode and file format choices that cut size without hurting OCR, and when to rescan instead. A scanned PDF is too large when each page is stored as a big, high-resolution, full-color image. Rescanning at about 300 dpi in grayscale, with normal compression, usually cuts the file to a fraction of its size while keeping text sharp enough for OCR. If the original scan is blurry or skewed, rescan instead of shrinking. ## Why is a scanned PDF too large when a downloaded one is tiny? A scanned PDF is too large because every page is a photograph of paper, while a downloaded statement is a page of stored text. Text takes a few bytes per character. A picture of the same page stores millions of pixels, each with its own shade, whether that pixel is ink, paper or a coffee stain. ### Where the megabytes come from The size of each scanned page depends on three settings chosen at scan time: - **Resolution**, measured in dots per inch (dpi). Doubling the dpi quadruples the pixels, because it doubles both width and height. - **Color mode.** Full color stores three channels per pixel, grayscale stores one, and black and white stores a single on-or-off bit. - **Compression.** The scanner or app can save each image at full size or squeeze it, with stronger compression trading detail for space. A scanner left on high resolution, full color and light compression can produce pages of several megabytes each. Multiply by a year of statements and the file runs to hundreds of megabytes. ### Why the size matters for conversion A big file takes longer to upload, may exceed a converter's limit, and takes longer to process. Large scans also make spreadsheets downstream feel heavy if images get carried into the workbook. On the spreadsheet side, our guide to [fixing Excel when it is slow with many rows](/blog/excel-slow-with-many-rows) covers what happens once the data lands. ## Scan resolution for a scanned PDF that stays readable Scan resolution sets how much detail each page keeps, and for text recognition about 300 dpi is the usual target. The [Tesseract OCR documentation on improving quality](https://tesseract-ocr.github.io/tessdoc/ImproveQuality.html) says that engine works best on images of at least 300 dpi. Going much higher mostly adds size, while dropping far lower starts to blur small digits. ### Matching resolution to the print Bank statements often use small type for amounts and dates. At 300 dpi, that type usually stays crisp. At 150 dpi, thin digits such as 1 and 7 can lose their edges. At 600 dpi, the page is four times the pixels of 300 dpi, and the extra detail rarely changes what OCR reads on ordinary printed text. ### The fine print exception Very small print, such as footnotes or tiny reference numbers, may benefit from a higher setting. If you need those details, scan just that page higher rather than the whole stack. ## Color, grayscale or black and white for scanned statement pages Color mode is the setting that usually saves the most size on scanned statement pages. Most statements are printed in dark ink on light paper, so grayscale captures everything that matters for text at a fraction of full color's weight. Pure black and white is smaller still, but it carries a risk on pale or uneven pages. ### Grayscale as the default Grayscale keeps shades between black and white, which helps OCR separate faint print from paper texture. It drops the color channels, which a black-ink statement does not need. For most statement scans, it is the best balance of size and accuracy. ### When black and white goes wrong Black and white forces every pixel to be either ink or paper. On a page with a shaded background, a light gray table band or faded print, the scanner's cut-off can erase thin strokes or turn a shaded row into a black bar. The Tesseract documentation notes that its own conversion to black and white can struggle when the page background is of uneven darkness. Test one page before you scan a stack this way. ### When to keep color Keep color only when color carries meaning, for example red amounts that mark debits, or highlights a client added on purpose. Even then, consider scanning that page in color and the rest in grayscale. ## Image compression and file format inside the scanned PDF Image compression decides how much space each scanned page takes inside the PDF. Most scanner apps offer a quality slider or a choice of formats. Moderate compression at a sensible resolution gives small files with clean text. Heavy compression creates blocky smudges around characters, which is exactly where OCR looks for shapes. ### Settings compared The table below compares typical choices for one statement page. The sizes are rough examples to show direction, not measurements of any scanner. | Resolution | Color mode | Compression | Example size per page | Effect on OCR | | --- | --- | --- | --- | --- | | 600 dpi | Color | Light | 6 MB | Sharp, but slow to upload | | 300 dpi | Color | Medium | 1.2 MB | Sharp, heavier than needed | | 300 dpi | Grayscale | Medium | 0.4 MB | Sharp, good default | | 300 dpi | Black and white | Standard | 0.1 MB | Good on clean pages, risky on shaded ones | | 150 dpi | Grayscale | Heavy | 0.08 MB | Small digits blur, more misreads | Read the table by column. Resolution and color mode do most of the work. Compression is the last setting to turn, and the first one to go wrong when pushed too far. ### Searchable PDF output Some scanner apps can run OCR on the device and save a searchable PDF with a hidden text layer. That adds a little size but can make conversion faster, because the text is already there. Check the result by searching for a word you can see. ## Shrinking an existing scanned PDF without rescanning Shrinking an existing scanned PDF works when the scan was good but saved with oversized settings. A PDF editor or online compressor can resample the page images to a lower resolution, convert them to grayscale, and apply stronger compression. The goal is to remove pixels nobody needs, not detail the converter relies on. ### Steps that usually help 1. Delete blank pages, separator sheets and backs of single-sided pages. 2. Reduce image resolution to around 300 dpi if it is higher. 3. Convert color pages to grayscale where color carries no meaning. 4. Apply medium compression, not maximum. 5. Open the result at 100 percent zoom and read a row of amounts. If digits look fuzzy, undo the last step. ### A worked example Here is an example with invented figures. A client sends a 12-page scan of 84 MB, about 7 MB per page. The pages are straight and sharp, so shrinking is worth trying. Removing four blank backs leaves eight pages. Resampling to 300 dpi and switching to grayscale brings the file to roughly 5 MB. A read of one row of amounts at 100 percent zoom shows clean digits, so the smaller copy is the one to upload. ### Watch for one-way changes Shrinking cannot add detail that was never captured. Each round of compression also loses a little more, so work from the original file each time instead of shrinking an already-shrunk copy. ## When to rescan instead of shrinking a scanned PDF Rescan instead of shrinking when the existing scanned PDF is the problem, not just its size. If pages are crooked, cropped, blurred, shadowed or out of order, a smaller copy only shrinks the same faults. When you still have the paper, a fresh scan with the right settings is faster than fixing a bad file. ### A rescan-or-shrink checklist Rescan if any of these are true: - Text near the edges is cut off. - Pages are visibly tilted. - Some pages are photographed with shadows or glare. - Digits look soft even at the original size. - The file is a phone photo stitched into a PDF. Shrink if all of these are true: - Every page is straight and complete. - Text is sharp at 100 percent zoom. - The only problem is the file size. ### A better source than any scan Before rescanning, check whether the bank offers the statement as a download in online banking. A downloaded statement is a digital PDF, small and already text, and it avoids recognition errors altogether. The same rule holds for other paperwork: when an online account offers the document as a download, take that over a scan, whether it is a statement or a [medical bill you want in Excel](/blog/medical-bill-pdf-to-excel). ## Do smaller scanned PDFs convert faster? Smaller scanned PDFs upload faster, and they often process somewhat faster too, but size is not the only thing that sets conversion time. The converter still has to recognize the text on every page. Page count, print quality and layout all matter. A small file of blurry pages can take longer to read well than a larger, sharp one. ### Speed and accuracy pull in different directions Cutting resolution or using heavy compression can make a file quick to send and slow to get right, because blurred characters lead to misreads you then fix by hand. The guide on [speed versus accuracy in PDF conversion](/blog/speed-vs-accuracy-in-pdf-conversion) looks at where that balance sits. ### What to expect from a scanned conversion FastBankConvert, which is getting ready to open, is built to read a typical digital statement in seconds; scanned statements take longer. It is designed to tell you before download if balances do not add up, which is where a misread digit from a poor scan would show. You can join through [Get started](/get-started) to hear when trial conversions open.