ERF to SVG Converter

ERF to SVG Converter

ERF is a camera RAW photograph: an unprocessed grid of sensor readings from an Epson R-D1 rangefinder. SVG is an XML document describing shapes — paths, curves and solid fills — with no pixels in it at all. Nothing inside an ERF is a path, so this is not a re-wrap; it is a trace, in which the picture is redrawn as flat colour regions. That works on a logo or a stencil and posterises an ordinary photograph. This page documents both formats precisely, shows the reliable two-step route, and is direct about when a raster target is the correct answer instead.

How to Convert ERF to SVG

1. Upload Your ERF File: Drag the .erf onto the page or click "+ Add Files" to browse. Several files can be queued and traced with identical settings.
2. Set Number Precision: Open Advanced Options and move the "Number precision" slider — it runs from 1 to 10 and sits at 6. Positions 9 and 10 abort the conversion and return no file, because the slider sends the same value as the colour precision, which is only defined for 1 to 8. It is the only tracing control on this page, and the in-app note recommends 4 to 6 for most images.
3. Consider Rendering to PNG First: The tracer consumes decoded raster images, and a camera RAW is a sensor mosaic that has to be demosaiced, white-balanced and tone-curved before it is one. Running ERF to PNG yourself and feeding that PNG to the PNG to SVG converter puts the exposure, the crop and the pixel size under your control — same tracer, same slider, an input you have actually looked at.
4. Convert and Download: Click Convert and save the .svg. Files are uploaded over an encrypted connection, processed on our servers, and deleted automatically after a few hours — no sign-up, no watermark.

ERF Format at a Glance

SVG Format at a Glance

What Number Precision Controls

The slider is usually described as a coordinate rounding setting. In our pipeline it is handed to the tracer twice — once as colour precision and once as path precision — so a single value governs two unrelated parts of the trace. Colour precision is the number of significant bits kept per RGB channel before neighbouring pixels are grouped into regions; lower values merge more shades into one flat shape. Path precision is the number of decimal places written into each coordinate; lower values shorten the XML but can visibly nudge small edges.

The practical consequences: at 1 to 3 you get poster-like blocks and the smallest file, which suits two-tone logos and silhouettes. At 4 to 6 distinct flat colours stay separate with crisp, compact paths — the usual sweet spot, and the range the in-app note recommends. At 7 and 8 near-identical shades are preserved as separate regions, coordinate strings grow long, and the file can balloon. Raising the value never recovers detail the trace did not find; on a photograph it simply multiplies the number of shapes.

Treat 8 as the real top of the slider even though it travels to 10. Because the same value is sent as the colour precision, and colour precision is only defined for 1 to 8 significant bits per RGB channel, positions 9 and 10 stop the run with an out-of-range error rather than producing a finer trace. That mismatch between the control's range and the engine's is a live footgun, independent of anything else on this page.

When a Raster Target Is the Right Answer

Almost every ERF is a real photograph, and tracing is the wrong operation for one. The R-D1's 6.1-megapixel frame holds millions of subtly different pixels; the tracer must quantise them into a small palette of solid fills, so skies band, skin and foliage flatten, and the SVG often ends up larger than a JPG of the same shot because a busy image forces thousands of separate paths. Trace an ERF only when the frame happens to contain a simple high-contrast graphic — a sign photographed off a wall, a stencil, a scanned diagram. Otherwise use ERF to PNG for a lossless raster, ERF to JPG for something small and shareable, or ERF to TIFF for a print and archive master. Whatever you choose, keep the .erf itself: it is the only copy that still carries the unprocessed sensor data.

Frequently Asked Questions

Does this embed my ERF inside an SVG or genuinely vectorize it?

It vectorizes. The tracer detects colour regions and writes them out as real SVG paths and curves; it does not base64-encode the photo into an <image> element. That is why the output scales without pixellating for suitable artwork, and equally why a photograph comes back reinterpreted as flat shapes rather than preserved.

Why does a rendered PNG trace better than the RAW it came from?

Because tracing operates on decoded pixels, and an ERF is not decoded pixels — it is a TIFF/EP container holding a bit-packed sensor mosaic that still needs demosaicing, white balance and a tone curve before it is a viewable image at all. Whoever performs that development picks the exposure, the colour and the noise floor the tracer then sees. Running ERF to PNG yourself puts those decisions in your hands and, just as usefully, puts the render in front of your eyes: you can catch an exposure that would flatten three shadow tones into one shapeless region, and downscale the frame so the tracer follows the subject rather than sensor grain.

What value should I set Number precision to?

For flat artwork, 4 to 6 covers nearly every case — the range the tool itself recommends. Drop to 1 to 3 when the image is genuinely two-tone and you want the smallest possible file. Go above 7 only when near-identical shades must remain separate regions, and expect the file to grow quickly, because the value raises colour precision and path precision together.

Will the SVG keep my EXIF, white balance or the RAW sensor data?

No. SVG has no metadata block equivalent to a RAW file's, and tracing discards everything that is not shape and fill. Capture date, camera and lens, ISO, shutter speed and the RAW white-balance values are all absent from the output. If you need an image file that still carries capture metadata, ERF to TIFF is the right target.

Why is my traced SVG bigger than the original photo file?

Because a detailed image forces one path per colour region, and a photographic source can produce tens of thousands of them. All that coordinate text adds up, easily past the size of a compressed raster of the same picture. Vectorization saves bytes on simple artwork, never on photographs. Lowering Number precision merges regions and trims the file, but the real fix is choosing a raster target.

Can I get my photograph back out of the SVG afterwards?

Not meaningfully. The SVG contains freshly generated paths that approximate what the tracer saw, so rendering it back to pixels returns the posterised approximation rather than the original frame. Tracing is a one-way transformation, which is another reason to keep the .erf as your master.

Do design applications open the resulting SVG cleanly?

Yes — SVG is a W3C standard rendered natively by every current browser and imported by Illustrator, Inkscape, Figma, Affinity Designer and most other vector editors. The caveat is path count rather than file size: a photographic trace can emit tens of thousands of objects, and editors build one object per path, so they can become sluggish. Lowering Number precision reduces the path count directly.

Is my ERF kept private while it is converted?

Your file is uploaded over an encrypted connection, traced on our servers, and deleted automatically after a few hours along with the SVG it produced. Nothing is shared or made public, and there is no sign-up and no watermark.