Initializing... drag & drop files here
Supports: ODT
PNG is the right target when a page is mostly type: it has no block artefacts around letterforms and it can carry a transparent background. But two of its settings on this page behave differently from what most people assume — the default preset is not lossless, and the background is not transparent unless you ask for it. This walk-through covers the four steps, then explains exactly what those two controls do and how to fix the results people most often complain about.
The Quality Preset does not just turn a dial; at the top setting it changes which encoder runs at all.
| Quality Preset | What actually happens | Colours in the output | Alpha channel |
|---|---|---|---|
| Highest | True-colour PNG written directly, with maximum deflate effort | Up to 16.7 million | Preserved |
| Very High (default) | Palette optimiser targeting quality 95 | Up to 256 | Preserved |
| High / Medium / Low / Very Low / Lowest | Same palette optimiser, progressively looser targets (90, 80, 70, 60, 50) | Up to 256 | Preserved |
For a normal text document this distinction rarely matters — a page of black type on white needs only a handful of greys for antialiasing, so 256 colours is far more than enough and the palette version is dramatically smaller. It starts to matter the moment the page contains a photograph, a gradient fill, a soft drop shadow, or a chart with subtle colour ramps. Those need thousands of shades, and squeezing them into 256 produces visible banding.
PNG is the only image target on this site that can hold a transparent page background, but you have to select it. The Image Transparency list opens on White, which flattens the page onto solid white before saving.
| Image Transparency setting | Effect on the finished PNG |
|---|---|
| Unchanged | Alpha preserved — everywhere the document has no content, the PNG is transparent |
| White (what the list opens on) | Page composited onto solid white |
| Black, plus 23 other named colours | Page composited onto that colour |
"Unchanged" is genuinely useful when you want to drop a page onto a coloured slide or a patterned web background without a white rectangle around it. It is not useful if the destination re-flattens the image anyway — most chat apps, many email clients, and every JPEG-only upload form will composite it onto their own background, usually white, sometimes black.
Rendering depends on the fonts that were available at conversion time. A document that references an unusual commercial typeface will be laid out with a metrically similar substitute, which can shift line breaks and repaginate long documents — the visual result is close but not guaranteed identical to what you see on your own machine. Embedding the fonts in the ODT before uploading avoids this, and checking the page count afterwards is a cheap sanity test. If pagination has to be exactly right, or if the text needs to stay selectable and searchable, convert to a fixed-layout document instead: ODT to PDF keeps a real text layer and packs every page into one file.
Only at the Highest quality preset. Every other preset, including the Very High default, runs a palette optimiser that reduces the page to at most 256 colours before writing the PNG — a lossy step, even though PNG's own compression is lossless. It is a very good trade for text pages and a poor one for photographs. If you need pixel-exact output, choose Highest explicitly; it is not what the page opens on.
Because the Image Transparency control on this page opens on White rather than on "Unchanged". The document is composited onto white before the PNG is written. Select "Unchanged" from that list and the alpha channel survives, giving you a page with genuinely transparent margins.
Ten, numbered in page order and downloadable individually or together. A PNG file holds exactly one image; the format has no concept of stacked pages. If you want all pages inside one file, use ODT to PDF. The animated variant, APNG, exists for motion rather than pagination and is not a document container either.
Around 300 DPI is the working floor for reliable recognition of ordinary body text, and 400-600 DPI is worth using for small print, footnotes, or degraded source material. Going beyond 600 DPI rarely improves recognition and inflates files quickly, because pixel count scales with the square of DPI. Also set Quality Preset to Highest for OCR work — palette reduction is exactly the kind of subtle edge damage that recognition engines are sensitive to.
PNG compresses by finding repetition; JPEG compresses by discarding detail your eye is unlikely to notice. A page of flat white with black type is enormously repetitive, so PNG does well and often beats JPEG. A page dominated by a photograph has almost no exact repetition, so PNG has very little to work with while JPEG thrives. Match the format to the page rather than expecting one to win everywhere.
None of it. Rasterising keeps the visible appearance and discards everything else — characters, font files, styles, tracked changes, comments, author fields and document properties. Search, copy-paste and screen readers all see a picture. That is often a feature when you are sharing a page publicly, but it is not a redaction tool: anything visible on the page is still perfectly visible in the image.
Not into the original. You can bundle the images into a PDF with an image-to-PDF conversion, and you can OCR them to recover approximate text, but the styles, tables, lists and section structure are gone for good. Keep the ODT as your master copy — the images are a derivative, not a replacement.
There is no sign-up and no watermark. Your file is uploaded over an encrypted TLS connection, processed on our servers, and deleted automatically a few hours after conversion — never shared or made public. The realistic constraint is upload size and connection speed rather than a fixed page cap. If the finished PNGs are what is too large, either drop the Conversion Quality a step or run them through the PNG compressor.