PEF to SVG Converter

Converting a Pentax PEF to SVG: Read This First

SVG is not a picture format in the way PEF is. It is an XML language for describing shapes — W3C published SVG 1.1 (Second Edition) as a Recommendation on 16 August 2011 — so getting from a photograph to SVG means tracing: an algorithm finds regions of similar colour and draws vector outlines around them. Two things follow, and both matter before you upload anything. First, the trace can only be as good as the raster it is given, and a Pentax .pef is not a raster — it is sensor data that has to be demosaiced, white-balanced and tone-curved into an image before there are any colour boundaries to find, so rendering it to PNG yourself is what puts that development under your control. Second, even with a perfect render, a continuous-tone photograph traces into a posterised illustration rather than a faithful picture. This page explains the working route and the exact limits.

How to Convert PEF to SVG

1. Render the Pentax Raw to PNG First: A raw file has to be developed into RGB pixels before anything can trace it, and doing that step yourself is what lets you check the exposure and colour before committing. Run the file through PEF to PNG and keep the resulting .png.
2. Upload the PNG to the SVG Tracer: Take that PNG to PNG to SVG. PNG is the best intermediate here because it is lossless, so you are not asking the tracer to find edges in JPEG compression artefacts.
3. Set Number Precision: One slider controls the trace, labelled "Number precision", with a range of 1 to 10 and a default of 6. Lower values simplify the output into fewer, coarser paths; higher values keep more detail and produce a larger file. The on-page note recommends 4 to 6, and there is a hard reason not to go past 8 — see the table below.
4. Convert and Download: Click Convert and save the .svg, then open it in Illustrator, Inkscape, Figma or any vector editor to recolour or reshape individual regions. Files are uploaded over an encrypted connection, processed on our servers, and deleted automatically after a few hours — no sign-up, no watermark.

What the Number Precision Slider Does

That single control is sent to the tracer as both its colour precision and its path precision, which is why one slider changes two very different things at once.

Positions 9 and 10 are therefore not usable. If you want the most detail the tracer can give, 8 is the ceiling.

What Makes a Good Tracer Input

Renaming a file changes nothing here, in either direction: the bytes inside a .pef are still a sensor mosaic whatever extension you give them, and a PNG renamed to .pef is still PNG bytes. Only a genuine render turns one into the other.

Common Errors and How to Fix Them

• "No image file found at specified input path" — the tracer was handed something it could not open as an image at all. The usual cause is a file renamed rather than converted. Run a real render with PEF to PNG and trace that.
• Nothing downloads and no error is obvious — the most common cause after an unreadable input is a Number precision of 9 or 10, which the tracer rejects as an invalid colour precision. Drop it to 8 or below.
• The SVG is enormous — you traced a photograph. Continuous tone needs thousands of paths to approximate what was smooth gradient. Lower the precision, or crop and downscale the PNG before tracing.
• The result looks like a poster, not a photo — that is what tracing does, not a bug. Vector paths are filled with flat colour; there is no equivalent of a photographic gradient in a traced result.
• Edges look ragged — you traced a JPEG whose compression artefacts the algorithm mistook for detail. Retrace from a PNG.

When to Use a Different Target Entirely

If you want to see the Pentax photograph, SVG is the wrong destination and no setting will fix that. PEF to JPG gives a small, universally viewable image, PEF to PNG keeps full colour losslessly and is also the correct intermediate for tracing, and PEF to TIFF gives an editable raster master for print work — set its Compression Type to LZW there, because the default is lossy JPEG. Reserve SVG for content that was already flat-shaded — a logo photographed off a wall, a scanned line drawing, a high-contrast graphic — where the trace produces clean, infinitely scalable shapes instead of banded approximations.

Frequently Asked Questions

Why render the PEF to PNG rather than trace the raw directly?

Because a raw file has to be developed either way, and the only question is who chooses how. A .pef records one filtered brightness value per photosite, not RGB pixels; demosaicing, white balance and a tone curve all have to be applied before there are colour regions to cluster. Rendering it yourself with PEF to PNG means you get to look at the result, correct an exposure that would otherwise clip the shadows into a single flat black region, and pick a sensible pixel size — three decisions that affect the trace far more than any tracing setting does.

Would renaming my PEF to .png work as a shortcut?

No, and it makes things worse. An extension is a label, not a conversion: renaming a raw to .png hands a PNG decoder a file full of TIFF-based sensor data, which it has no way to interpret. The reverse is equally pointless — a genuine PNG renamed to .pef is still PNG bytes wearing the wrong name. There is no rename that substitutes for an actual render.

Why does one slider control both colour and path precision?

Because the page exposes a single "Number precision" value and passes it to the tracer as both parameters. It is a deliberate simplification — two sliders would need explaining, and for the flat graphics that trace well the two settings tend to move together anyway. The side effect is the hard ceiling at 8: the tracer accepts a path precision of 9 or 10 but rejects a colour precision above 8, so the combined value cannot go that high.

What does the tracer actually do to my image?

It scans the raster, groups adjacent pixels into regions of similar colour, and emits a vector path around each region filled with a single flat colour. Nothing is embedded — the original pixels are discarded, and the SVG contains only shape definitions. That is why a two-colour logo traces beautifully and a landscape does not: the logo already is a small number of flat regions.

Why does my photograph come out banded and posterised?

Because flat fill is the only tool a traced SVG has. A Pentax sensor records 12 or 14 bits per channel of continuous tone across skies, skin and shadow; the tracer collapses those gradients into a handful of discrete colour regions with visible boundaries between them. Lowering the precision makes the bands larger and the file smaller; raising it makes them finer but never actually smooth. If you want photographic tone, you want a raster format.

Will the SVG be smaller than my original Pentax raw?

For a flat graphic, dramatically — a traced logo is often a few kilobytes against tens of megabytes of raw. For a photograph, often not: a detailed scene can need thousands of paths, each with coordinate data, and the resulting XML can exceed a good JPEG of the same picture while looking worse. Precision is the main lever; scaling the PNG down before tracing is the other.

Can I edit the traced SVG afterwards?

Yes, and that is the main reason to trace at all. Because the output is a set of vector paths, you can open it in Illustrator, Inkscape, Figma or any vector editor and recolour, reshape, delete or rescale individual regions with no loss of sharpness at any size. That is impossible with a raster export, however high its resolution.

Does tracing keep any of the raw's colour depth or metadata?

No. A PEF's value is its high-bit-depth sensor data and its EXIF and Pentax maker notes, and tracing discards both — the output is XML shape data with flat fill colours. Nothing about white balance, exposure or lens information survives. Keep the original .pef as your archive and treat the SVG as a derived graphic.

How are my uploads handled while this runs?

Every file is uploaded over an encrypted connection, processed on our servers, and deleted automatically after a few hours. There is no sign-up, no watermark, and files are never shared or made public. Because the working route is two steps, both the intermediate PNG and the traced SVG are covered by the same handling.