How to export pdf diagrams and charts as png images
- Step 1Trim to the diagram pages first (optional but smart) — If the diagram is buried in a long spec, run PDF Extract Pages to pull just the figure pages. Fewer pages also keeps a free-tier job under the 50-page cap.
- Step 2Open the converter and drop the PDF — Load PDF to PNG and drop the file. It auto-converts every page — no Convert button, no options panel.
- Step 3Let each page render at 144 DPI — pdf.js rasterises every page at the fixed 2x scale and encodes a lossless PNG. The resolution is constant; there is no DPI control.
- Step 4Collect the page-numbered PNGs — Each page downloads as
…-page-N.png. A multi-figure PDF triggers one download per page (~200 ms apart) — allow multiple downloads if prompted. - Step 5Embed in your docs — Drop the PNG into Confluence/Notion/MkDocs/GitHub. Add descriptive alt text for accessibility, since diagram labels are now pixels, not selectable text.
- Step 6If labels are too small, enlarge in source — Because DPI is fixed at 144, dense schematics with tiny callouts read better if you crop/enlarge the diagram in the source app and re-export the PDF before converting.
PNG vs JPG for technical diagrams
Both render at the same 2x (144 DPI). PNG wins for line art; JPG only helps when a figure is photographic (e.g. an embedded photo of hardware).
| Figure type | Use | Why |
|---|---|---|
| Flowchart / UML / ERD | PNG (this tool) | Lossless lines & text, no ringing |
| Architecture diagram | PNG (this tool) | Crisp boxes, arrows, labels |
| P&ID / wiring schematic | PNG (this tool) | Thin lines survive losslessly |
| Chart with text labels | PNG (this tool) | Axis text stays sharp |
| Photo of equipment / rendering | PDF to JPG | Smaller file, ringing irrelevant on photos |
Diagram readability at the fixed 144 DPI
Whether tiny labels stay legible depends on how much page area the diagram occupies. The tool can't raise DPI, so source layout is your lever.
| Situation | At 144 DPI | What to do |
|---|---|---|
| Diagram fills the page | Labels crisp and readable | Convert as-is |
| Diagram is half the page | Usually fine for screen | Convert; crop the PNG if needed |
| Small inset with 6pt callouts | Callouts get soft | Enlarge/crop in source, re-export PDF, then convert |
| Multi-diagram page | Each diagram smaller | Split diagrams onto separate pages first |
| Wide diagram on portrait page | Lots of whitespace | Rotate/landscape the source page before export |
File-size and page limits by tier
Real limits from the PDF tool family. The free tier blocks oversized files or page counts with an upgrade prompt before any rendering starts; everything runs locally in your browser either way.
| Tier | Max file size | Max pages per PDF | Files per job |
|---|---|---|---|
| Free | 2 MB | 50 | 1 |
| Pro | 50 MB | 500 | 5 |
| Pro + Media | 500 MB | 2,000 | 50 |
| Developer | 2 GB | 10,000 | 1 (unlimited batch) |
| Enterprise | Unlimited | Unlimited | Unlimited |
Cookbook
Real documentation workflows — getting diagrams from PDF into a wiki or docs site cleanly, with the right format choice.
Architecture diagram into Confluence
A single architecture page becomes a clean PNG you embed inline with alt text.
architecture.pdf (1 page) -> PDF to PNG -> architecture-page-1.png 1224 x 1584 px -> Confluence: Insert > Image > upload -> alt: "Service architecture: API gateway, 3 services, DB" Labels stay sharp; no JPEG halo around the boxes.
Spec with figures on pages 3, 7, 11
Pull only the figure pages, then convert, so you don't generate a PNG of every text page.
spec.pdf (40 pages, figures on 3, 7, 11) 1. PDF Extract Pages -> "3,7,11" -> figures.pdf (3 pages) 2. PDF to PNG -> figures-page-1/2/3.png Three clean diagram PNGs, no clutter, under the free 50-page cap.
Flowchart into a GitHub README
PNG embeds natively in Markdown and renders crisply at 100% on GitHub.
flow.pdf -> PDF to PNG -> flow-page-1.png README.md:  Lossless lines render clean on GitHub's white background.
Tiny callouts that won't read at 144 DPI
When the diagram is a small inset, fix it in source rather than fighting the fixed DPI.
Problem: schematic inset with 6pt labels -> soft after export Fix in source (Visio/Lucid/CAD): - crop to the diagram, scale it to fill an A4 page - re-export PDF - PDF to PNG -> now labels are large enough to read
Wide diagram, lots of empty margin
A landscape diagram on a portrait page wastes pixels on whitespace. Set the source page to landscape first.
wide-diagram.pdf (portrait, diagram across the top) -> set source page to landscape, re-export -> PDF to PNG -> diagram fills the frame, more detail per pixel -> or crop the exported PNG to the diagram bounds
Edge cases and what actually happens
Tiny labels are soft after export
Fixed 144 DPI limitThere is no DPI control, so a small diagram with 6pt callouts can blur. Enlarge or crop the diagram in the source app so it fills the page, re-export the PDF, then convert. Bigger source artwork = more pixels per label.
You wanted an editable diagram, not an image
Raster, by designPNG is a flat image — labels aren't selectable and the diagram can't be re-laid-out. To edit, recover the vector source (the original Visio/Lucid/CAD/draw.io file) or keep the PDF; rasterising is one-way.
Diagram spans two pages
Stitch manuallyThe tool exports one PNG per page, so a two-page diagram becomes two PNGs. Stitch them in an image editor, or fix the source so the diagram fits one page before exporting.
White background in the docs theme
Matches defaultPNGs are rendered on white, which matches most docs/wiki themes. On a dark-themed site the white box will show — cut out the white (see the transparent-PNG guide) or invert the diagram in source for dark mode.
Free tier and the spec is over 50 pages
Blocked — upgradeA 51+ page PDF is blocked on free. Extract just the figure pages first (almost always a handful), which keeps you under the cap and avoids generating PNGs of text pages anyway.
Free tier and the file is over 2 MB
Blocked — upgradeVector-heavy specs can exceed 2 MB. Extracting the figure pages usually shrinks the file under the free cap; otherwise Pro raises the limit to 50 MB. Nothing is uploaded — the check is local.
Encrypted spec PDF
Render failsAn encrypted PDF can't be rendered without the password. Remove it with PDF Remove Password first, then convert.
Embedded fonts render as boxes
Rare, font fallbackIf a diagram uses an unusual embedded font, the renderer normally uses the embedded glyphs; missing/corrupt font data can fall back and shift labels. Re-export the PDF with fonts embedded if labels look wrong.
Photographic figure looks heavy as PNG
Use JPGA page that's mostly a photo of equipment produces a large PNG. For photographic figures, PDF to JPG gives a much smaller file with no visible quality cost.
Multi-figure PDF only saved one PNG
Allow multiple downloadsEach figure page is a separate download fired ~200 ms apart. The browser may block the extras — click 'Allow' on the multiple-downloads prompt so they all save.
Frequently asked questions
Why is PNG better than JPG for diagrams?
PNG is lossless, so thin lines, arrowheads, and small labels stay sharp. JPEG adds compression ringing around hard edges that makes diagrams look fuzzy. The only time JPG wins is a photographic figure, where its smaller files outweigh artefacts on a photo.
What resolution do the diagram PNGs come out at?
A fixed 2x scale, i.e. 144 DPI — a Letter page is 1224x1584 px, A4 is 1190x1684 px. There's no DPI slider. For dense schematics with tiny callouts, enlarge the diagram in the source app and re-export the PDF before converting.
Can I export only the pages that have diagrams?
Not within this tool — it converts every page. Use PDF Extract Pages to pull just the figure pages (e.g. '3,7,11') into a small PDF, then drop that here. You'll get one clean PNG per figure and avoid PNGs of text pages.
Will the diagram labels still be searchable text?
No. PNG is a flat raster — labels become pixels, not selectable or searchable text. Add descriptive alt text when you embed the PNG for accessibility and search. If you need the text, extract it separately with PDF to Text.
How do I keep tiny callouts readable?
Because DPI is fixed at 144, the lever is source layout: crop or scale the diagram so it fills the page, then re-export the PDF and convert. A diagram that fills an A4 page gets far more pixels per label than a small inset.
Can I import the PNG into Confluence, Notion, or MkDocs?
Yes — PNG is natively supported as an inline image in Confluence, Notion, GitHub/GitLab, MkDocs, Docusaurus, and every major wiki and docs platform. No conversion needed; just upload or reference the file.
My diagram spans two pages — what happens?
You get two PNGs, one per page. Stitch them together in an image editor, or adjust the source so the whole diagram fits a single page (landscape orientation often helps) before exporting.
Is my schematic uploaded to a server?
No. All rendering happens in your browser and the file is never uploaded. Only the pdf.js worker script loads from a CDN. This is important for proprietary or internal-only engineering documents.
The diagram has a white box on my dark-themed site — fix?
PNGs render on white (there's no transparency option here). On a dark theme, cut out the white in an editor — the transparent-PNG guide covers the exact steps — or produce a dark-mode version of the diagram in source.
What if an embedded font renders incorrectly?
The renderer uses the PDF's embedded fonts; problems usually mean the font wasn't fully embedded. Re-export the source PDF with 'embed all fonts' enabled, then convert again. This fixes shifted or boxed labels.
Can I get a higher-DPI export for a poster-size diagram?
Not from this tool — 144 DPI is fixed. For large-format output, keep the vector PDF (it scales infinitely) and send that to print, or enlarge the diagram in source before converting if you specifically need a bigger raster.
How big a spec can I convert at once?
Free allows 2 MB and 50 pages; Pro 50 MB / 500 pages; higher tiers more. For long specs, extracting just the figure pages first is both faster and keeps you under the free caps — you rarely need PNGs of the text pages.
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.