Initializing... drag & drop files here
Supports: ERF
ERF is the raw format of the Epson R-D1, the first digital rangefinder ever built, and of the two cameras that followed it. Fewer than a handful of bodies ever wrote the format and Epson left the camera business afterwards, so ERF sits in an awkward spot: the files are ordinary TIFF/EP underneath, but very little consumer software bothers to claim them. Converting one to MPEG-2 produces a short silent video in an MPEG program stream — useful when the destination is a DVD authoring application, a broadcast ingest workflow or a piece of playback hardware from the same era as the camera.
| Property | Value |
|---|---|
| Full name | Epson RAW Format |
| Container | TIFF/EP — the file begins with the big-endian TIFF magic MM\0* |
| MIME type | image/x-epson-erf |
| Sensor data | Bit-packed 12-bit CFA data, stored with compression tag 32769 |
| Cameras | Epson R-D1 (2004), R-D1s (2006), R-D1x (2009) — all 6.1 MP APS-C rangefinders |
| Adopted by anyone else? | No. Epson left the camera business and no other maker used ERF |
| Decoded size, R-D1 sample | 3040 × 2024 pixels |
| Property | Value |
|---|---|
| Standard | ITU-T H.262, first edition July 1995; published jointly as ISO/IEC 13818-2 |
| Current edition | H.262 (02/12), still the in-force text |
| Container written here | MPEG program stream (ffmpeg's mpeg muxer) under an .mpeg2 name |
| Streams in the output | One video stream, no audio |
| Frame rate | 1 frame per second, with B-frames disabled |
| Where it is still required | DVD-Video, SVCD, DVB and ATSC ingest, older signage and in-flight players |
.erf onto the page or click "+ Add Files". Only .erf is accepted here. The file is uploaded over an encrypted connection, processed on our servers, and deleted automatically after a few hours — no sign-up, no watermark, never shared or made public.Two of the groups on this page look like quality controls and are not. MPEG-2 has no constant-quality mode, so the Quality Preset falls through to a bitrate calculation based on the source video's bitrate — and a still photograph does not have one. Constant Quality and Constraint Quality are both rewritten as a plain bitrate request before the encoder sees them, and that request is then replaced by a floor derived from the output frame size. Everything therefore comes back to resolution. Encoding the R-D1 test file at its full 3040 × 2024 gave a 376,832-byte five-second clip; the same frame at 766 × 576 came back at 51,200 bytes. The R-D1's 6-megapixel CCD is modest by modern standards, so the gap here is smaller than it would be with a 20-megapixel raw — but resolution is still the only control on the page that moves the number.
Not because the format is exotic — ERF is a TIFF/EP file and starts with the standard MM\0* TIFF magic — but because it is rare. Three camera models across five years, from a manufacturer who then exited the market, is not enough volume for most photo applications to justify writing and testing a decoder. Libraries built specifically for raw decoding do handle it; consumer viewers and phone galleries generally do not.
They should. All three bodies are built around the same 6.1-megapixel APS-C CCD and all three write ERF, and there is nothing model-specific in the decoding path here — the file is handed to a general raw decoder, not to per-body logic. Our own test file came from an R-D1. If a particular file refuses to decode, that is far more likely to be damage or truncation than a model difference.
At "Keep original" it matches the decoded raw. Our R-D1 sample decodes to 3040 × 2024, which is the full sensor area rather than the rounded 6.1-megapixel figure. That is well under the 4096-pixel ceiling the encoding path applies, so nothing is scaled down unless you ask for it.
It is a valid bitstream, but it is far outside the profiles and levels that hardware decoders implement — Main Profile at High Level tops out at 1920 × 1152. Software players such as VLC will handle it; a DVD player, a broadcast decoder or an old signage box will not. If the file has to play on hardware, pick 720 × 576 for PAL or 720 × 480 for NTSC using "Width x Height".
Because the source is a photograph. The Audio Codec group is hidden on image-to-video conversions and the output contains a single video stream. That also means the resulting .mpeg2 cannot be fed into an audio extractor afterwards — a file with no audio stream fails outright rather than producing silence.
Not unless you force it. "Keep original" and the height presets both preserve the raw's own aspect ratio, so a 3:2 frame stays 3:2. Choosing a fixed 16:9 resolution instead adds pillarbox bars down the sides rather than stretching the picture, and those bars are filled with the Background Color, which starts on White.
Yes. Queue the files, leave Merge strategy on "Merge images", and each raw is held for the Image Duration in turn; ten frames at four seconds gives a forty-second clip. "Video per image" is the alternative when you want the files kept separate.
Almost certainly, unless something in your chain specifically demands MPEG-2. ERF to MP4 produces a far smaller file that plays in browsers and on phones. If you want the photograph rather than a video, ERF to JPG or ERF to TIFF skips the video encode altogether and keeps every pixel the R-D1's CCD recorded.