Initializing... drag & drop files here
Supports: MOS
MOS is the raw file that Leaf and Mamiya medium-format digital backs write. ExifTool's supported-formats table identifies the type as "Creo Leaf Mosaic" and classifies it as TIFF-based: the undemosaiced sensor mosaic sits inside a TIFF wrapper rather than in a finished, viewable picture. WebP is Google's web image format, built on the RIFF container and readable by every current browser.
Converting MOS to WebP is therefore a development step, not a repackaging. Our servers demosaic the capture, apply a tone curve to turn linear sensor levels into a normal-looking image, and encode the result as WebP. That is why the output opens instantly in a browser while the .mos never will — and also why the conversion only runs one way.
| Property | Value |
|---|---|
| Type identifier | Creo Leaf Mosaic (ExifTool supported-file-types table) |
| Underlying container | TIFF |
| Payload | Undemosaiced sensor mosaic plus capture metadata |
| Written by | Leaf digital backs — Aptus 17, 22, 54S, 65, 65S, 75, 75S; Aptus-II 5 through 12R; AFi and AFi-II; Valeo; Credo — and the Mamiya ZD (models named in LibRaw 0.22's supported-camera list) |
| Alpha channel | No — a raw capture has no transparency |
| Native browser support | None; browsers do not decode raw mosaics |
| Editing latitude | Full — white balance, exposure and highlight recovery are still adjustable |
| Best for | Archiving the original capture and re-editing in a raw developer |
| Property | Value |
|---|---|
| Container | RIFF (Google's WebP container specification) |
| Lossy mode | "8-bit Y'CbCr 4:2:0", per Google's WebP FAQ |
| Lossless mode | RGBA, per the same FAQ |
| Alpha channel | Yes; Google puts the lossless transparency cost at "just 22% additional bytes" |
| Size versus JPEG | Google measures lossy WebP at "25-34% smaller than comparable JPEG images at equivalent SSIM quality index" |
| Size versus PNG | Google measures lossless WebP at "26% smaller in size compared to PNGs" |
| Hard limits | Canvas width x height at most 2^32 - 1; the whole file at most 4 GiB minus 2 bytes |
| Native browser support | Chrome 32+, Firefox 65+, Edge 18+, Safari 16+ on desktop, iOS Safari 14+ (caniuse) |
| Best for | Publishing, sharing and storing a finished frame at web sizes |
.mos onto the page or click "Add Files" to browse. Several captures can be queued and developed with the same settings in one batch.A raw file records what the sensor measured; a WebP records what the picture should look like. Everything a raw developer would let you revisit later — white balance, exposure, highlight rolloff — is decided during this conversion and baked into the output pixels.
The bit depth goes with it. Google's WebP FAQ states that lossy WebP "works exclusively with an 8-bit Y'CbCr 4:2:0 (often called YUV420) image format" and that lossless WebP "works exclusively with the RGBA format", so whatever depth your back recorded, the WebP carries 8 bits per channel. Lossy WebP additionally subsamples the colour channels to 4:2:0; the Lossless setting avoids that but not the drop to 8 bits.
None of this is a reason to avoid WebP — it is the normal cost of publishing a finished picture. It is a reason to keep the .mos. If you want a developed file that still holds 16-bit data for further editing, render to MOS to TIFF instead; for a photo that opens anywhere at all, MOS to JPG is the safer target.
No, and no tool can do it. The MOS holds an undemosaiced sensor mosaic; the WebP holds a demosaiced, tone-curved, 8-bit picture. The conversion discards the raw measurement in favour of one interpretation of it, so there is nothing left to reverse-engineer. Archive the .mos alongside the WebP if the capture matters.
It preserves the render exactly — the WebP will match the developed image pixel for pixel with no compression artefacts. It does not preserve the raw capture, because the demosaic and tone curve happen before the encoder ever sees the data, and because lossless WebP is still an 8-bit RGBA format. Think of Lossless as "no second-generation loss on top of the development", not "no loss at all".
Because the pixel dimensions come straight from the sensor. Medium-format backs produce very large frames, and "Keep original" resolution hands all of those pixels to the encoder. Drop the Quality Preset a step, or scale the render with Resolution Percentage or Preset Resolutions — the presets set the target height and let the width follow your image's aspect ratio, so a portrait and a landscape frame at the same preset end up different widths.
Per caniuse, WebP decodes natively in Chrome from version 32, Firefox from 65, Edge from 18, desktop Safari from 16 and iOS Safari from 14. That covers essentially every browser still receiving updates. Older Safari builds on macOS Catalina and earlier are the realistic exception; JPG is the fallback if you must support them.
Treat the WebP as a picture, not as a record of the shoot. WebP's container can carry EXIF and XMP chunks, but a raw file's most useful metadata — the maker-note block describing the back, the as-shot white balance, the raw levels — is specific to the raw format and has no equivalent slot in a finished image. If the capture metadata matters for your archive, keep the .mos or render to MOS to TIFF, which retains a far richer metadata structure.
Lossy, in almost every case. It is the "No (Recommended)" default, it is the mode Google measures at 25-34% smaller than an equivalent JPEG, and at the Very High preset the difference is not visible at normal viewing sizes. Lossless is worth choosing when the render will be re-edited or re-encoded downstream and you do not want a second generation of compression stacked on the first.
Because file size scales with pixel count much faster than it scales with quality. Halving both dimensions removes three-quarters of the pixels; dropping one quality step only changes how many bits each pixel costs. For a medium-format frame headed for a web page, resize first and adjust quality second — it is the difference between a slightly softer huge image and a sharp appropriately-sized one.
They are uploaded over an encrypted connection, developed on our servers and deleted automatically after a few hours. Nothing is shared or made public, there is no sign-up and nothing is watermarked. The practical constraint on a large medium-format capture is upload time rather than the conversion itself. If you already have a WebP and only want it smaller, compress WebP reduces the file without changing the format.