Initializing... drag & drop files here
Supports: MRW
MRW is the raw format Minolta and Konica Minolta wrote between roughly 2001 and 2006 — DiMAGE 5/7/7i/7Hi and A1/A2/A200, then the Dynax/Maxxum 5D and 7D. It is not a picture yet. It is a mosaic of single-colour sensor readings plus the camera's own white-balance multipliers, and something has to develop it before anything can display it.
A .jfif is that developed picture, written as a JPEG. Not "like a JPEG" — the identical encoded bytes you would get from a .jpg, under a four-character extension. You land on this page when an importer, an archive spec, or an old ingest script asks for .jfif by name.
An MRW is not a TIFF derivative, which is why generic TIFF readers choke on it. The first bytes of the sample used for the measurements below tell the story directly.
| Property | MRW source | JFIF output |
|---|---|---|
| File signature | 00 4D 52 4D — a null byte then MRM, followed by a PRD block |
FF D8 then an FF E1 Exif segment |
| Structure | Proprietary block container (PRD/TTW/WBG/RIF/PAD), not TIFF/EP | JPEG marker segments per ITU-T T.81 |
| Published as | Never a public standard; Minolta internal | JFIF 1.02 (1992), later ITU-T T.871 (2011) and ISO/IEC 10918-5 (2013) |
| Pixel data | One colour per photosite, 12-bit class, undeveloped | Full-colour 8-bit sRGB, DCT-compressed |
| Colour decisions | Deferred — white balance stored as multipliers | Baked in permanently |
| Cameras that wrote it | DiMAGE 5/7/7i/7Hi, A1/A2/A200, Dynax/Maxxum 5D and 7D | n/a |
| Superseded by | Sony ARW, after Sony absorbed the Konica Minolta SLR line in 2006 | Nothing; JPEG is still the floor of image interchange |
| Opens anywhere | No — needs a raw-aware decoder | Yes, in every browser and image editor |
.mrw onto the page or click "+ Add Files". Several frames can be queued and they all run with the same settings..jfif. Files are uploaded over an encrypted connection, developed and encoded on our servers, and deleted automatically after a few hours — no sign-up, no watermark.Every preset maps to one JPEG quality number handed to the encoder. These are the shipped values, not estimates. The byte column is a measured re-encode of a 3016 x 2008 Dynax/Maxxum-era MRW through the same encoder settings the converter uses.
| Quality Preset | JPEG quality | Chroma sampling | Measured output (3016 x 2008 frame) |
|---|---|---|---|
| Highest | 100 | 4:4:4 | 6.85 MB |
| Very High (Recommended) | 95 | 4:4:4 | 2.44 MB |
| High | 90 | 4:4:4 | 1.67 MB |
| Medium | 80 | 4:2:0 | 0.83 MB |
| Low | 70 | 4:2:0 | 0.66 MB |
| Very Low | 60 | 4:2:0 | 0.54 MB |
| Lowest | 50 | 4:2:0 | 0.47 MB |
The chroma column is the part people miss. At quality 90 and above the encoder keeps full-resolution colour (4:4:4); at 80 and below it halves colour resolution in both directions (4:2:0). That is a much bigger visual step than the quality number suggests on saturated edges — red text on a dark background, for example. If a frame has fine coloured detail, the jump from Medium to High buys more than one preset step normally would.
This is the one result that surprises people, so it is worth stating plainly: the JFIF this page produces carries essentially no camera EXIF. Running the ALPHA-7 DIGITAL sample through the exact encoder invocation the converter uses, the output held an orientation tag and a resolution field and nothing else — no Make, no Model, no ISO, no shutter speed, no aperture, no focal length, no capture date. The same encoder preserves all of those when the input is already a JPEG, which is how we know it is the raw decoding stage and not the JPEG writer dropping them: the raw loader publishes shot data under its own private field names that the JPEG writer does not know how to re-emit.
Practical consequence: keep the original .mrw. It is the only copy of the shot data, and it is also the only copy that can still be re-developed with different white balance later.
Yes. JFIF is the interchange convention that defines how JPEG-compressed data is packaged — the marker segments, the pixel-density fields, the colour conventions decoders expect. It appeared as JFIF 1.02 in 1992 and was formalised much later as ITU-T T.871 (2011) and ISO/IEC 10918-5 (2013). A .jfif and a .jpg written at the same quality are byte-for-byte the same picture. If an application refuses the file, renaming it to .jpg is a complete fix, or run it through JFIF to JPG.
The camera's own. The raw decoder is configured to use the white-balance multipliers the body recorded at capture time, then demosaics to 8-bit sRGB. So the JFIF should look close to the JPEG that camera would have written itself, not to a neutral or auto-corrected interpretation. If the shot was taken under mixed or unusual lighting and the camera guessed wrong, that wrong guess is what gets baked in — develop the MRW in RawTherapee or darktable first and export from there if colour accuracy is the point.
Yes, unless you change it. Image resolution opens on Keep original, and a still-image target has no frame-size ceiling to run into — the 3016 x 2008 sample came out 3016 x 2008. That is different from the video targets on this site, where a still is capped at 4096 pixels on the long edge before encoding. If you want it smaller, Preset Resolutions set the height and let width follow the source aspect ratio, while Width x Height forces both.
Because the encoder writes an Exif APP1 segment for the orientation flag, and a JPEG carrying Exif conventionally does not also carry the JFIF APP0 segment. The file is still an ordinary baseline JPEG and every decoder reads it; the "JFIF" in the filename describes the family, not a marker that must be present. Nothing about this affects display, colour, or compatibility.
Not as a JFIF — JPEG is lossy by construction, and even quality 100 is a lossy DCT encode. For a lossless developed image use MRW to TIFF, which is the sensible archival target, or MRW to PNG for a lossless web image. Note that the PNG path applies palette quantization at most presets, so pick the highest preset there if lossless is the requirement.
No. An MRW holds roughly 12 bits per photosite of headroom, which is what lets you pull back a blown sky or shift white balance long after the shot. The JFIF is 8 bits per channel with the tone curve already applied. Highlight and shadow recovery that would have worked on the raw file will band or clip on the JPEG. Set the look you want before exporting, and keep the .mrw as the master.
Different Minolta and Konica Minolta bodies wrote slightly different MRW variants, and the format was never publicly specified, so support was always reverse-engineered per camera. Add a decade of software dropping legacy raw types after Minolta left the camera business in 2006 and you get exactly this: one viewer opens the DiMAGE files, another opens the 7D files, neither opens both. Converting the whole archive to one format is the durable fix.
No, and they are not interchangeable. Sony bought the Konica Minolta SLR line in 2006 and the first Alpha body was essentially a 5D design, so the hardware lineage is real — but ARW is a TIFF-based container and MRW is Minolta's own block format with an MRM signature. Tools that read current Sony raw files will usually not read MRW.
Yes. Queue as many .mrw files as you like; they convert with the same quality and resolution settings and come back individually or as a ZIP. The realistic constraint on a big archive is upload size and time over your connection rather than anything in the settings. If you would rather have .jpg filenames at the end, MRW to JPG writes the identical image data under the more portable extension.
They are uploaded over an encrypted connection, decoded and encoded on our servers, and both the uploaded MRW and the JFIF it produced are deleted automatically after a few hours. Nothing is shared or made public, there is no account to create, and no watermark is added.