PDF to PNG or JPG Guide: Page Rendering, PPI, Quality and Export Limits
Admin ·
PDF to image renders selected visible PDF pages into individual PNG or JPG files. Choose one supported local PDF, enter physical page positions in the desired order, select Page resolution and format, then export and inspect the generated images. Each repeated page receives its own result and filename.
This is complete-page rasterization rather than extraction of original embedded pictures or creation of editable PDF content. PNG and JPG both use a white page background. Selected PPI determines the raster grid, while encoded image density metadata, JPEG quality, preview detail and source-resource limits have separate meanings.
Choose one PDF and export complete page images
Use Choose PDF or drop one nonempty supported local PDF into the workspace. Enter the desired physical positions in Pages, choose Image format and Page resolution, optionally enable Grayscale, then press Export page images. Download the completed images after checking their displayed dimensions and appearance.
Defaults are PNG, 150 PPI, JPEG quality 90 and Grayscale off. Page resolution offers 72, 96, 150, 200 and 300 PPI. Only JPG displays the JPEG quality slider. The chooser does not automatically validate the source page count or range; export can be enabled before those checks run.
- Identify the complete source and selected-occurrence limits before exporting.
- Choose a useful raster size and inspect text or diagrams at the resulting pixels.
- Keep the original PDF when selectable text, vectors or interactive features matter.
MDN: local files and the File APIPDF.js: visible page bounds, units, rotation and rendering
Rendering pages differs from extracting original pictures
Each result represents the complete visible PDF page area, including text, vector drawings and images as rendered pixels. It is not a separately extracted original JPEG or PNG resource. Crop bounds, physical page units and existing rotation contribute to the rendered page viewport.
The image does not retain selectable PDF text, vector paths, clickable links, interactive form controls or the original encoded bytes of embedded pictures. A scanned page remains an image rendered inside a new page image; this operation does not perform OCR. It has no text-editor, asset-extraction, TIFF or JPEG 2000 output mode.
PDF.js: visible page bounds, units, rotation and renderingMDN: PNG and JPEG format characteristics
Entered physical page order and repetitions are preserved
Pages accepts ASCII decimal positions, commas and ascending hyphen ranges. Blank or whitespace-only text selects every source page. 3,1,3 requests the third, first and third physical pages in that order. Printed numbers, Roman front matter and PDF page labels are not the selection syntax.
Repeated positions create separate output occurrences and count again toward page, pixel and encoded-byte totals. Range text has a 5000-code-unit UI limit. Invalid syntax, descending ranges, page 0 or positions outside the source can fail during export rather than being reflected by a disabled export button.
| Selection | Result behavior |
|---|---|
| Blank selection | Request every source page; the separate 50-occurrence ceiling still applies |
3,1,3 | Render three occurrences in entered order |
1-3,5 | Render positions 1,2,3,5 in that order |
| Repeated page entry | Create another named result and count its pixels/bytes again |
| Printed labels or Roman numerals | Use physical one-based source positions instead |
Page resolution scales the visible PDF viewport
The renderer uses the selected PPI as a scale of PPI divided by 72 on the PDF page viewport. Crop bounds, the PDF UserUnit and existing page rotation affect its dimensions; width and height are rounded upward to whole pixels. Increasing PPI increases the raster grid, not the detail available in a low-resolution source scan.
For a visible one-inch square, nominal outputs are 72 × 72, 150 × 150 and 300 × 300 pixels at the corresponding PPI. Doubling PPI approximately quadruples pixel area before rounding. This workspace has no export rotation, flip, crop or automatic downscale-to-limit control; reduce Page resolution yourself if the resulting grid exceeds a limit.
| Selected PPI | Nominal one-inch square | Meaning |
|---|---|---|
72 | 72 × 72 pixels | One pixel per PDF point when UserUnit is 1 |
96 | 96 × 96 pixels | Larger raster grid |
150 | 150 × 150 pixels | Default export setting |
200 | 200 × 200 pixels | More pixels and larger area |
300 | 300 × 300 pixels | Highest offered setting; limits still apply |
Selected PPI is not a rewritten image-file DPI tag
Page resolution sets the page-to-pixel rendering scale. The app then encodes the canvas without separately rewriting PNG or JPEG density metadata. Canvas.toBlob documentation specifies 96 DPI resolution metadata for formats that support it, so the chosen 150 or 300 PPI setting is not a promise of a matching stored density tag.
Check actual pixel dimensions when placing the exported image. If another program chooses a print size from density metadata, explicitly set the intended physical size or use a workflow that deliberately manages density tags. A default encoded tag does not reduce the raster pixels already rendered at your chosen page scale.
PNG and JPG encode the rendered canvas differently
PNG losslessly encodes the rendered canvas pixels; that does not make the raster identical to original PDF resources, fonts, vectors or profiles. JPG uses lossy encoding and can be useful for photographic page content, while text and line art can show compression artifacts. Compare the actual results at the chosen dimensions.
Only JPG shows JPEG quality, from 10 to 100, default 90. The value requests an encoder quality, not a universal perceptual score or byte-size target. Quality 100 does not make JPEG lossless or recover missing source detail. Switching to PNG hides the slider but remembers its value for a later switch back to JPG.
| Setting | Current effect |
|---|---|
| PNG | Lossless encoding of the already rendered canvas |
| JPG | Lossy image encoding; inspect compression artifacts |
| JPEG quality | Requested 10–100 setting, visible only for JPG |
Quality 100 | Does not guarantee lossless encoding or restored source detail |
| Switch JPG → PNG → JPG | Retain the remembered JPEG quality value |
MDN: canvas encoding, quality and density metadataMDN: PNG and JPEG format characteristics
Both formats render with a white page background
The app explicitly renders onto a white background for PNG and JPG. Although PNG as a format supports alpha transparency, this page-export workspace has no transparent-background setting. Blank page areas are rendered as opaque white rather than preserving a transparent PDF backdrop.
Visible annotation rendering is enabled, but this is not a guarantee of every form widget, XFA document, scripted appearance, media object or reader-specific effect. Interactive behavior is not carried into the raster image. Inspect the intended visible content before sharing an export, particularly when a form or annotation is important.
PDF.js: visible page bounds, units, rotation and renderingMDN: PNG and JPEG format characteristics
Grayscale changes rendered RGB pixels, not print profiles
Grayscale processes the rendered display RGB values using a weighted sum, rounds the result and sets the red, green and blue channels to that same value. It preserves alpha; the preceding white page render supplies the opaque background. The setting is available for PNG and JPG.
This is not a dedicated CMYK, ICC print-profile or linear-light luminance conversion. Color-management and print-production requirements need their own controlled workflow. When Grayscale is off, inspect the actual rendered colors; rendering a PDF into an image does not certify perfect reproduction of every source color space.
Source pages, output occurrences and raster budgets are separate
Choose one nonempty PDF up to 50 MB with 1–300 complete source pages. Export at most 50 selected occurrences per job, including repetitions. A supported 300-page source fails a blank selection because blank requests all 300; select at most 50 intended occurrences. A smaller selection cannot admit a source with more than 300 pages.
Each output image must remain within 12 megapixels and 8192 pixels on either side. The sum of all output-image areas is capped at 150 megapixels and encoded image bytes at 100 MB, before ZIP packaging. Exceeding a limit fails the export without automatic resolution reduction or partial result cards.
| Boundary | Current rule |
|---|---|
| Source bytes | Up to 50 MB: 50000000 bytes |
| Complete source pages | 1–300, before subset selection |
| Selected occurrences | At most 50, including repeats |
| Every rendered image | At most 12000000 pixels; each side at most 8192 pixels |
| Combined image area | At most 150000000 pixels across results |
| Combined encoded image bytes | At most 100 MB: 100000000 bytes before ZIP packaging |
An oversized image on an unselected page can block rendering
Source inspection checks embedded image resources across the document, not just the selected pages. An embedded image above 24 megapixels or above 12000 pixels on either side can prevent rendering. A valid embedded 12001 × 1 image on a different source page can therefore block an export of page 1.
Reducing PPI or selecting fewer pages helps output-raster budgets but cannot bypass this whole-source resource guard. It differs from the page-copy tools, which can sometimes copy pages when previews are unavailable. This raster exporter needs supported rendering; it does not offer a copy-only fallback or an original-picture extraction mode.
Source and rendering errors fail the complete export
Source parsing can refuse encrypted, damaged or unsupported-header files and detected signature fields or markers, including recognized unused fields. The workspace has no password entry, unlocking or cryptographic signature-validation control. These are source-editing and rendering policies rather than a complete PDF-conformance audit.
The strict renderer can also reject broken image content or other diagnostics rather than silently returning a page with missing content. A file that another viewer partially displays is not guaranteed to export here. Old results are cleared before processing, and a later selected-page failure exposes no partial new result list.
Clear file and Choose another PDF retain settings
Clear file removes the source, results and messages while preserving Pages text, Image format, Page resolution, JPEG quality and Grayscale. Choosing another valid PDF also keeps those settings. Check retained ranges against the replacement document; a formerly valid position can be out of range in the new source.
Choosing multiple, empty or oversized files reports an error, clears previous results/messages and retains the already selected source. Canceling the picker without choosing a file leaves the workspace alone. Settings edits invalidate old results and require another Export page images action; they do not automatically regenerate files.
Source preview detail does not set export resolution
Preview source pages browses the original document independently of the selected export order. Its 800, 1600 and 2400 pixel longest-side display settings do not alter exported image pixels, JPEG quality or stored output files. Changing export settings keeps a mounted source preview while clearing prior results.
While export is busy, the source preview is unmounted; after processing or cancellation it remounts as a fresh closed preview. Replacing a valid source updates its preview page/count state, while open/detail state can persist. Avoid interpreting a preview zoom or a remembered export range as an automatic resize or source change.
Output names preserve occurrence order and source positions
Names use a three-digit occurrence prefix and the physical source page position. For a PNG selection 3,1,3, the files are 001-page-3.png, 002-page-1.png and 003-page-3.png. JPG uses the .jpg suffix. Ordinal names keep repeated pages distinct; they do not use the source basename or translate with the page language.
Download each image individually. When more than one result exists, Download all as ZIP creates toolmellow-results.zip with stored entries; it does not recompress JPEG or PNG payloads. ZIP wrapper overhead sits outside the 100 MB combined encoded-image cap. One result has no ZIP action.
Resize images or deliberately assemble a new image-based PDF
Each result offers Resize image, passing that one generated image through same-tab memory to the localized image-resizer tool under its limits. When there are 2–20 image results, Make one PDF passes the collection to the image-to-PDF workflow. That batch action is absent for one result and for 21–50 results.
Putting rendered page images back into a PDF does not restore original selectable text, vectors, links, forms or embedded-resource bytes. These are separate deliberate handoffs, not an automatic editable-PDF round trip, cloud save or hosted-file conversion API. Keep the original PDF for workflows that need its document features.
Cancel retains the source and settings
The whole export has a nominal 120-second timeout. Cancel aborts processing while keeping the input and settings. No partial new result set is committed after cancellation, and previously downloaded files are not recalled. Fewer selected pages or a lower resolution can help output-workload limits when the source itself is supported.
PDF source inspection and PDF.js parsing use support workers. The app invokes page rendering on a main-thread canvas, then performs grayscale processing and requests browser canvas encoding. It does not promise that the whole export runs in a worker. Ordinary hosting, vendor font/CMap/ICC/Wasm assets and analytics requests are separate from local PDF-byte processing; local work does not establish anonymity or zero logging.
MDN: Web Worker messaging and lifecycleMDN: stopping a workerMDN: canvas encoding, quality and density metadata
Embed the tool with visible ToolMellow attribution
Use Add this tool to your website on the localized page to copy or download the supplied HTML snippet. Retain the iframe, sizing behavior and visible ToolMellow backlink with its existing link attributes. Choose the desired language and allow room for settings, source preview and multiple result cards.
Visitors choose their own supported local PDF under the same source, occurrence, pixel and byte limits. This snippet is not a hosted-file conversion API. Test the intended frame layout and browsers, inspect real generated images and keep the attribution visible. Refreshing or leaving ends the transient workspace, so save needed downloads separately.
MDN: local files and the File APIMDN: Web Worker messaging and lifecycle
Common questions
How do I export PDF pages to PNG or JPG?
Choose one supported local PDF, enter physical page positions or leave Pages blank for all, select PNG or JPG and one offered PPI value, optionally adjust JPEG quality or Grayscale, then press Export page images and inspect the generated files.
Does this extract original embedded pictures or render complete pages?
It renders complete visible pages into raster images. It does not extract original embedded JPEG/PNG bytes or retain editable PDF text, vector paths or interactive content.
Why can a supported 300-page PDF fail a blank page selection?
Blank means all source pages. The complete source can have up to 300 pages, but one export is limited to 50 selected occurrences. Select a smaller set; a source above 300 pages remains unsupported.
Are page order and repeated page entries preserved?
Yes. ASCII comma entries and ascending ranges preserve entered physical order and repeats. Every repeat counts again toward the 50-occurrence, total-pixel and encoded-byte budgets and receives a distinct ordinal filename.
How does Page resolution PPI change exported pixels?
The selected PPI scales the visible PDF viewport by PPI/72, accounting for crop bounds, page units and rotation; each dimension rounds upward to whole pixels. Higher PPI adds raster pixels rather than recovering detail missing from the source.
Does selected PPI set the image file DPI metadata?
The app does not rewrite density tags to the chosen PPI. Canvas encoding documentation specifies 96 DPI metadata for supporting formats. Check actual pixel dimensions and explicitly set intended print size when another program interprets density tags.
Does PNG keep transparent PDF page backgrounds?
No. Both PNG and JPG page rendering use a white background. PNG format alpha capability does not create a transparent-background control in this workspace.
How do PNG and JPG quality settings differ?
PNG losslessly encodes rendered canvas pixels. JPG uses lossy encoding with a requested 10–100 quality setting, default 90. JPEG quality 100 is not lossless or a detail-recovery guarantee, and PNG is not preservation of original PDF vectors/resources.
Does grayscale convert print profiles or CMYK?
No dedicated print-profile or CMYK conversion is provided. Grayscale uses weighted rendered display RGB values, rounds them and sets RGB equally while preserving alpha; inspect results for your intended workflow.
Can lower PPI bypass oversized embedded-image guards?
No. Embedded resources anywhere in the source above 24 MP or 12000 pixels on either side can block rendering before output selection. Lower PPI can help output raster limits, not bypass source-resource refusal.
Do Clear file and Choose another PDF reset page ranges and settings?
No. Both retain ranges, format, PPI, remembered JPEG quality and Grayscale. Check those settings for a new file. A rejected replacement keeps the existing source while clearing old results and reporting the error.
Can I embed PDF to image with a visible credit link?
Yes. Copy or download the provided HTML iframe snippet, preserving its visible ToolMellow attribution/backlink and sizing. Visitors choose supported local PDFs; the snippet is not a hosted-file conversion API.
What happens if I cancel or the export times out?
Cancel retains source and settings but commits no partial new image results. The whole export has a nominal 120-second timeout. Already downloaded files remain on your device.
Does the output preserve every annotation or interactive form?
PNG and JPG render visible annotations, but there is no universal guarantee for every form, widget, XFA or scripted appearance. The result is a raster image with a white background rather than interactive PDF content.
Sources and further reading
- PDF.js: visible page bounds, units, rotation and rendering
- PDF.js: viewport and canvas rendering examples
- MDN: canvas encoding, quality and density metadata
- MDN: PNG and JPEG format characteristics
- W3C PDF17: labels and physical page positions
- MDN: local files and the File API
- MDN: Web Worker messaging and lifecycle
- MDN: stopping a worker