MillFile

Convert RAF to JPG

RAF is Fujifilm's own raw format, and it is unusual under the hood. Drop one here for a full-resolution JPG, decoded on your device - nothing is uploaded.

Quality:

Selected: Standard

Why RAF behaves differently from most raw files

Fujifilm did not build RAF on TIFF like most raw formats. It wraps the sensor data in an Exif-based structure with a custom layout, and that is why RAF has a history of upsetting converters: tools that assume the usual container read the wrong offsets and report strange dimensions or a broken preview. LibRaw has a dedicated path for RAF, and that is the decoder running in this tab.

Fujifilm's film simulations - Pro Neg Hi, Velvia, Astia, Classic Chrome - are stored alongside the raw data rather than baked into it. So a RAF converted here gives you the picture with the white balance recorded in the file and no simulation applied: a neutral full-resolution render rather than the look of the film mode you were previewing in the camera.

Nothing is uploaded. The RAW file is decoded in this tab by LibRaw compiled to WebAssembly — no upload step, no server copy, no account.

Questions about converting RAF to JPG

Why does RAF break other converters?

Because it is not a TIFF. Fujifilm wraps the sensor data in an Exif-based structure with its own layout, so a decoder that assumes the common raw container reads the wrong offsets and comes back with implausible dimensions or an unusable preview. Decoders written for RAF handle it correctly; the one here is.

Will my film simulation be applied?

No. Velvia, Pro Neg Hi and the rest are stored next to the raw data rather than inside it, and choosing one is a rendering decision made afterwards. This page renders with the white balance recorded in the file, so what comes back is a neutral image rather than the look you were previewing.

Does it handle GFX medium format files?

Yes - GFX bodies write RAF as well, and the decoder is the same one. Expect a longer wait than an APS-C frame, because a medium format sensor has far more pixels to demosaic, and expect it to need more memory while it does.

Does this use the embedded preview instead?

No. As on the other format pages here, the frame is decoded from the sensor data, which is why it takes a couple of seconds instead of being instant. In exchange you get the full frame rather than the smaller JPEG Fujifilm stores for playback on the camera.