You go to compress a PDF and a settings panel presents you with terms like DPI, downsampling, JBIG2, JPEG2000, and lossless versus lossy. These are not everyday vocabulary. They are technical terms from image processing and compression engineering. But they control exactly what happens to your PDF when you hit the compress button, and understanding them at a basic level helps you choose settings that produce a readable document instead of a pixelated mess.
These terms break down into two categories: resolution controls, which determine how much image detail is kept, and compression methods, which determine how the surviving detail is stored. Once you understand what each term controls, the compression settings panel becomes a set of informed choices rather than a guessing game.
DPI: The Resolution Knob That Controls Everything
DPI stands for Dots Per Inch. It measures how many pixels are packed into each inch of an image. A 300 DPI scan of a letter-sized page contains roughly 8.5 million pixels, 2550 pixels wide by 3300 pixels tall. A 150 DPI scan of the same page contains roughly 2.1 million pixels, 1275 by 1650. A 72 DPI version contains about 0.5 million pixels, 612 by 792. The jump from 72 DPI to 300 DPI represents 17 times more image data to store, which is why scan resolution is the dominant factor in PDF file size.
For on-screen viewing, 150 DPI is the practical sweet spot. A typical computer monitor displays 92 to 220 pixels per inch, so a 150 DPI image has enough resolution to look sharp without wasting storage on pixels the screen cannot show. For printing, 300 DPI is the standard because the human eye can perceive detail up to roughly 300 dots per inch on paper held at reading distance. When a compression tool asks you to choose a DPI target, it is asking: what is the lowest resolution that is still acceptable for how this document will be used?
Downsampling: How the Resolution Reduction Actually Happens
Downsampling is the process of reducing an image's resolution. If you tell the compression tool to downsample all images to 150 DPI, it examines each image in the PDF, calculates its current effective resolution, and if that resolution is higher than 150 DPI, it reduces the pixel count until the image is at 150 DPI. Images that are already at 150 DPI or below are left alone.
There are three common downsampling algorithms, and they produce noticeably different results. Average downsampling takes the average color value of a block of pixels and replaces the entire block with that single value. It is fast but tends to blur fine details. Subsampling takes a single pixel from the center of each block and discards the rest. It is faster than averaging but can miss important details that happen to fall between the sampled pixels. Bicubic downsampling uses a weighted average of surrounding pixels to calculate each new pixel value. It is the slowest method but produces the smoothest, most accurate result. For text-heavy scanned documents, bicubic downsampling preserves the sharp edges of letterforms better than the alternatives. For photographs, the difference between methods is less pronounced.
Compression Methods: JPEG, JPEG2000, JBIG2, and Lossless
After the image has been downsampled to the target resolution, it still needs to be encoded into bytes. The choice of compression method determines how efficiently those pixels are stored and whether any image quality is sacrificed in the process.
| Method | Lossy or Lossless | Best For | File Size Impact |
|---|---|---|---|
| JPEG | Lossy | Color photos, scanned pages with images | High compression; may show artifacts at low quality |
| JPEG2000 | Lossy or Lossless | High-quality photos, documents needing both quality and small size | Better compression than JPEG at same quality |
| JBIG2 | Lossy or Lossless | Black-and-white scanned text pages | Extreme compression; up to 10x smaller than JPEG for B&W |
| Flate/Deflate | Lossless | Graphics, screenshots, line art | Moderate compression; no quality loss |
| CCITT G4 | Lossless | Fax documents, pure black-and-white pages | Very good for 1-bit images; poor for anything else |
The most important distinction in the table is lossy versus lossless. Lossy compression, used by JPEG and some modes of JPEG2000 and JBIG2, permanently discards image detail to achieve smaller file sizes. The discarded detail cannot be recovered. Lossless compression, used by Flate and the lossless modes of JPEG2000 and JBIG2, finds mathematical redundancies in the pixel data and encodes them more efficiently without discarding any information. The original image can be reconstructed perfectly from a losslessly compressed file.
JBIG2: The Specialist for Black-and-White Documents
JBIG2 deserves special attention because it is uniquely suited to the most common type of scanned PDF Compression document: black text on a white background. JBIG2 works by identifying recurring patterns on a page and storing each pattern once with references to its locations, rather than storing each occurrence separately. On a page of typed text, every instance of the letter "e" in the same font and size is nearly identical. JBIG2 stores one copy of that "e" and a list of where it appears. This pattern-matching approach compresses typed text far more aggressively than JPEG, which treats the page as a generic image.
There is an important caveat with JBIG2 in its lossy mode. The pattern matching algorithm can sometimes confuse similar-looking characters. An "r" followed by an "n" might be stored as a single "m" if the algorithm incorrectly merges the pattern. This substitution, called a JBIG2 artifact, changes the actual text content of the document. For archival documents where text accuracy is paramount, use JBIG2 in lossless mode or choose a different compression method entirely. The file size will be larger, but the content will be guaranteed accurate.
What Compression Settings to Use for Common Document Types
For a black-and-white scanned contract or report, bicubic downsampling to 150 DPI with JBIG2 lossless compression produces a dramatically smaller file with perfectly preserved text. File size reduction of 70% to 85% is typical for this combination on text-heavy scans.
For a color brochure or photo book, bicubic downsampling to 150 DPI with JPEG2000 at a quality setting of medium to high balances image quality and file size. File size reduction of 50% to 65% is typical.
For a mixed document containing both text pages and photo pages, the ideal approach is to apply different settings to different image types. Many PDF Quality compression tools, including WukongPDF, automatically detect whether each image is color, grayscale, or monochrome and apply the optimal compression method for each type. This per-image optimization produces better overall results than applying a single setting to every image in the document.
The next time a compression settings panel asks you to choose between JPEG2000 and JBIG2, or to set a downsampling DPI, you will know that these are not arbitrary technical switches. They are specific tools for specific jobs. Selecting the right tool for the document in front of you is the difference between a compressed PDF that looks indistinguishable from the original and one that looks like it was run through a fax machine from 1995.