How to compress high-resolution images in a pdf document
- Step 1Confirm the PDF is print-resolution imagery — If it came from InDesign, Illustrator, a photo-book exporter, or 300-DPI scans and is image-dominated, this is the right tool. If it's mostly text with a few images, lossy would needlessly fuzz the text — use lossless compress instead.
- Step 2Open the compressor and drop the PDF — Load it into PDF Compress (Aggressive). Rendering with pdf.js and re-assembly with pdf-lib happen in your browser — your high-res assets are never uploaded to us.
- Step 3Let it render and re-encode automatically — There is no DPI or quality field. The tool renders every page at the built-in
1.5×scale and re-encodes each page as a JPEG at the built-in quality (~0.65), then assembles the result — one pass, no dialog. - Step 4Give long, high-res documents time — Each high-resolution page is rasterised individually in the browser, which is memory- and time-intensive for a 100-page photo book. The tab stays responsive; on very large sets, splitting with PDF Split first keeps memory manageable.
- Step 5Inspect images at 100% on your target device — View the compressed file at actual size on the screen it's meant for. Screen fidelity at the built-in quality is good; fine gradients and small detail soften. If a key image is too soft, that's inherent — your print master remains the source of truth.
- Step 6Archive the master; publish the compressed copy — Lossy compression is irreversible and the flattened text can't be restored. Keep your full-resolution original for print and re-use; distribute only the screen-optimised copy.
What the tool controls — and what it doesn't
Setting expectations precisely: the browser tool applies fixed built-in settings and exposes no resolution or quality controls.
| Control you might expect | Available here? | Reality |
|---|---|---|
| Target DPI (72 / 150 / 300) | No | Pages render at a fixed built-in scale (1.5×); there is no DPI selector. |
| Per-image downsampling | No | The whole page is rasterised and JPEG'd; individual image objects aren't touched separately. |
| JPEG quality slider | No (web UI) | Built-in quality (~0.65) is applied automatically; quality is an engine/API parameter, not a page control. |
| Controlled resolution stepping | In target-size mode | The target-size workflow steps the render scale down (e.g. 1.5→1.25→1.0→0.85) to hit a byte target. |
| Keep vector text crisp | No | Text and vectors are rasterised into the page image — use lossless to preserve them. |
Right tool for a high-res source
Match the source to the tool; lossy is for image-dominated, screen-bound output.
| High-res source | Best tool | Why |
|---|---|---|
| Design export / photo book for screen | Lossy (this tool) | Whole-page JPEG collapses print-resolution imagery for screen. |
| Must hit an exact byte size | Target-size workflow | Only this mode steps resolution down to land under a target. |
| Mixed text + a few high-res photos | Lossless | Keeps text selectable; lossy would fuzz the text for little gain. |
| Colour images for a B&W deliverable | Grayscale then compress | Removing colour channels first makes the JPEG smaller. |
Input tier limits
Maximum input the in-browser compressor accepts. Local processing only; not a server quota.
| Tier | Max input file | Max pages | Files per run |
|---|---|---|---|
| Free | 2 MB | 50 | 1 |
| Pro | 50 MB | 500 | 5 |
| Pro Media | 500 MB | 2,000 | 50 |
Cookbook
High-resolution PDF scenarios and what the fixed-scale re-render produces.
InDesign print export, now for the website
A 16-page brochure exported at print resolution from InDesign is 28 MB — fine for the printer, far too heavy to embed on a web page. Re-rendering each page for screen drops it sharply.
Input: brochure-print.pdf 28 MB 16 pages @ ~300 DPI imagery Process: pages rendered at 1.5x, JPEG'd (~q0.65) Output: ~3-5 MB Use: web embed / email; keep the 28 MB print master
Camera-resolution photo book
A 40-page book built from full-resolution camera JPEGs is 120 MB. Each page is a photo, so lossy is ideal — but it's a big job for the browser.
Input: photobook.pdf 120 MB 40 full-res photo pages Process: each page rasterised + JPEG'd (slow, memory-heavy) Output: a screen-friendly fraction of the original Tip: on Pro Media (500 MB cap); split if the tab struggles
You actually need a specific byte size
A client portal wants the brochure under 4 MB exactly. The fixed-scale pass won't aim for a number — the target-size workflow steps resolution down to land under it.
Standard tool: produces one result at fixed scale/quality For an exact ceiling: /pdf-tools/guides/lossy-compress-pdf-under-1mb targetSizeKb: 4096 (4 MB) -> binary-searches quality; drops render scale if needed
Grayscale first for a mono print-on-demand
A colour design export destined for a black-and-white print-on-demand title. Converting to grayscale before compressing removes colour channels, so the page JPEGs are smaller.
Step 1: /pdf-tools/pdf-grayscale -> drops colour channels Step 2: /pdf-tools/pdf-compress-lossy -> JPEG the mono pages Result: smaller than compressing the colour version
Mixed report — don't lossy the whole thing
A research report with high-res figures but lots of body text. Flattening it makes 30 pages of text un-selectable to save a little. Use lossless to keep text and still trim.
Wrong: lossy -> figures shrink BUT all text becomes images
Right: /pdf-tools/pdf-compress-lossless
-> smaller, every page's text stays selectableEdge cases and what actually happens
Expecting to choose a target DPI (72 / 150 / 300)
By designThe browser tool has no DPI selector. It renders every page at a fixed built-in scale (1.5×) and JPEG-encodes it — you can't dial in 72 or 150 DPI here. For controlled resolution stepping, the target-size workflow lowers the render scale in a ladder to hit a byte target.
Expecting per-image downsampling that keeps text
not supportedUnlike a print-prep downsampler, this tool does not re-encode individual image objects while preserving the surrounding page. It rasterises the entire page to one JPEG, so text and vectors are flattened too. To keep text selectable, use lossless compress.
Selectable text is lost
trade-offWhole-page re-rendering bakes text into the image. That's inherent to lossy here. If the high-res PDF also needs selectable text, lossless is the only option — it can't shrink as aggressively, but it preserves the text layer.
You need a precise byte target
use target-sizeThe standard tool runs one fixed-scale, fixed-quality pass and won't aim at a number. For an exact ceiling, the target-size workflow binary-searches JPEG quality and steps the render scale down (1.5→1.25→1.0→0.85) until it lands under your target.
Very large, high-resolution document is slow
slow but worksRasterising many high-res pages in the browser is memory- and time-heavy. A 40-page camera-resolution photo book takes real time and can strain a phone. It completes — split with PDF Split first, or process on a desktop.
Free tier rejects the high-res file
size limitFree PDF input caps at 2 MB / 50 pages — most print-resolution exports exceed that immediately. Upgrade to Pro (50 MB / 500 pages) or Pro Media (500 MB / 2,000 pages), or split the source first so each part fits.
Vector logos and fine line art soften
ExpectedVectors are infinitely sharp until rasterised; here they're baked into the page bitmap and JPEG-encoded, so hard edges can show softening or ringing. Keep the vector master for print; use the compressed copy only where screen softness is acceptable.
Result not as small as a print downsampler gives
expected differenceA dedicated print-prep tool that downsamples each image to a chosen DPI may beat a single fixed-scale pass on some files. For a controlled size, use the target-size workflow; for true per-image control, downsample the images in image software before assembling the PDF.
Searchable OCR layer dropped
OCR lostIf a high-res scan carried a searchable OCR layer, re-rendering to images discards it. Re-run OCR on the compressed file if searchability matters, or keep the OCR'd original.
Frequently asked questions
Can I set the target DPI, like 72 or 150?
No — the in-browser tool has no DPI control. It renders every page at a fixed built-in scale (1.5×) and re-encodes each page as a JPEG at a built-in quality. There's no resolution dropdown. If you need controlled resolution reduction toward a specific file size, the target-size workflow steps the render scale down to hit a byte target.
Does it downsample only the images and keep the text sharp?
No. This is a whole-page re-render, not a selective image downsampler: the entire page (text and vectors included) is rasterised and JPEG-encoded. That's why it shrinks image-heavy PDFs so well, and why text on those pages stops being selectable. To keep text crisp and selectable, use lossless compress.
What DPI is appropriate for screen-only PDFs?
Conceptually, 72 DPI matches standard screens and ~150 DPI looks sharper on high-density (Retina) displays — but you can't set those here, because the tool has no DPI input. It uses a fixed render scale that produces good screen fidelity. For a specific resolution or size, use the target-size workflow, which controls the render scale via a ladder.
Should I keep the original high-resolution PDF?
Always. Lossy compression is irreversible, and the flattened text can't be restored. Use the compressed copy only for screen, web, or email distribution — print production needs the full-resolution master, which this tool does not modify.
Will charts and vector graphics be affected?
Yes — unlike a tool that only resamples raster images, this one rasterises everything on the page. Vector charts and logos are baked into the page bitmap and then JPEG-encoded, so their crisp edges soften. If vector sharpness matters, keep the original and use lossless, or export the vectors separately.
How much smaller will a print-resolution PDF get?
For genuinely image-dominated, 300-DPI sources, expect a large reduction — often a screen-friendly fraction of the original — because the print-resolution pixels are re-encoded at screen-appropriate fidelity. Exact results depend on page count and image content. A specific number isn't guaranteed by the fixed-scale pass; use target-size mode if you need one.
My PDF has both high-res figures and lots of text — what should I use?
Use lossless compress if the text must stay selectable; it trims the file while keeping every page's text intact. Only choose lossy if small size matters more than selectable text — it will flatten the whole document, including the text, into images.
Is my high-value design file uploaded anywhere?
No. Rendering and re-assembly run entirely in your browser; the file never reaches JAD's servers. That matters for unreleased design work and licensed photography. Only an anonymous usage counter is recorded when you're signed in.
Why is processing slow on my big photo book?
Each high-resolution page is rasterised individually in the browser, which is memory- and CPU-intensive at scale. A long photo book genuinely takes time and can strain a phone. It still finishes — split the document with PDF Split first, or run it on a desktop, to keep things smooth.
Can I get a result as small as a print-prep downsampler?
Sometimes a dedicated per-image downsampler beats a single fixed-scale pass. For a controlled outcome, use the target-size workflow, which lowers the render scale toward a byte target. For true per-image control, downsample images in image-editing software before building the PDF — this tool is built for fast, uniform whole-document reduction.
Will the high-res images go below the file's input limit after compression?
Compression happens after the file is loaded, so the input tier cap (Free 2 MB, Pro 50 MB, Pro Media 500 MB) applies to the original you upload, not the compressed output. If your print-resolution PDF exceeds the free cap, upgrade or split it first so it can be loaded and compressed.
Does compressing the result again shrink it more?
No. Re-encoding an already-lossy PDF just stacks JPEG artefacts for little size benefit. Compress the original once. If it isn't small enough, restart from the original with the target-size workflow and a specific kilobyte target rather than re-compressing the compressed file.
Privacy first
All PDF processing runs locally in your browser using PDF-lib and pdf.js. No file is ever uploaded — only metadata counters are saved for signed-in dashboard stats.