Initializing... drag & drop files here
Supports: CR2
A .cr2 is not a picture yet. Canon Raw version 2 is a TIFF-based container holding the values read straight off the sensor — typically 12- or 14-bit per photosite, compressed with lossless JPEG, behind a colour filter array that still has to be interpolated. Canon never published the specification, so every piece of software that opens one is working from reverse engineering. This page renders that raw file and re-encodes the result as AVIF, the AV1 Image File Format, which is the right target when the destination is a web page rather than an archive.
.cr2 onto the page or click "+ Add Files". Several Canon raws can be queued together, and they share one set of settings..avif. Files are uploaded over an encrypted connection, converted on our servers and deleted automatically a few hours later — no sign-up, no watermark.| Property | Value |
|---|---|
| Full name | Canon Raw version 2 |
| Structure | TIFF-based container; sensor data compressed with lossless JPEG (ITU-T T.81) |
| Specification | Proprietary and unpublished — third-party support is reverse-engineered |
| Sample depth | Typically 12- or 14-bit per photosite |
| Colour | Single-channel mosaic behind a colour filter array; needs demosaicing before display |
| Transparency | None — a camera frame is fully opaque |
| Era | Mid-2000s Canon DSLR and mirrorless bodies, until CR3 arrived in 2018 |
| Best for | Capture and archival, where exposure and white balance stay adjustable |
| Property | Value |
|---|---|
| Full name | AV1 Image File Format |
| Structure | AV1 intra-coded picture wrapped in the HEIF container (ISO/IEC 23008-12) |
| Published by | The Alliance for Open Media |
| Bit depth | 8, 10 or 12-bit, as chosen by the encoder |
| Transparency | Supported, via an alpha plane |
| Animation | Supported, as an image sequence |
| Browser support | Chrome 85+, Firefox 93+, Safari 16.4+ (iOS 16+), Edge 121+ — about 94.7% of tracked global traffic per caniuse |
| Best for | Delivering finished photographs on the web at small file sizes |
Every raw converter has to guess. The sensor recorded one colour per pixel site, so the missing two-thirds are interpolated; the neutral point has to be chosen; a tone curve has to be applied to map scene luminance onto a display. Canon's in-camera JPEG engine makes those decisions using the Picture Style you had set, which is why the preview on the back of the camera rarely matches what a converter produces from the same file.
Our render applies a standard demosaic and a neutral rendering rather than trying to imitate a particular Canon body. If you care about a specific look, do the interpretation yourself first: open the .cr2 in raw software, set white balance and exposure the way you want them, export, and convert that export. Once the AVIF exists, none of those decisions can be revisited — an AVIF is a rendered picture, not sensor data, no matter how many bits it carries.
Size drops sharply in this conversion for the same reason. A raw file stores the full dynamic range of every photosite because you might need it later; a delivery image stores only what a display will show, and AV1 intra coding is efficient at doing so. That is a good trade for a web page and a bad one for an archive, so keep the original .cr2 either way.
No, and no encoder setting changes that. The latitude in a raw file comes from having untouched sensor values with a wide dynamic range and no fixed white balance. Converting bakes a specific interpretation into the pixels. AVIF can carry 10- or 12-bit samples, which helps with banding in gradients, but extra bits describe the rendered image — they do not restore blown highlights or let you re-pick a colour temperature. Keep the .cr2 and treat the AVIF as the published copy.
Because that preview was a JPEG your camera rendered with its own Picture Style, sharpening and noise reduction, and it is embedded alongside the raw data rather than being the raw data. A converter starts from the sensor values and applies its own demosaic and neutral rendering. Differences in contrast, saturation and colour cast are expected. To match a look exactly, export from raw software first and convert that file.
Every current mainstream browser: Chrome from version 85, Firefox from 93, Safari from 16.4 on macOS and 16 on iOS, and Edge from 121. Desktop software is patchier — older versions of image editors and viewers that predate AVIF support will simply refuse the file. If your recipient is unknown, publish AVIF for the web and keep a CR2 to JPG copy for anything that has to open anywhere.
No. It is a delivery format: lossy by default, rendered, and dependent on decoder support that only became common in the 2020s. For archival, keep the original .cr2 files, which hold everything the sensor recorded. If you want an editable intermediate that opens in almost anything, CR2 to TIFF is the safer choice, at a much larger file size.
Do not count on it. A raw file has to be rendered to an intermediate image before it can be encoded as AVIF, and camera metadata does not reliably survive that step. If shot settings, GPS or lens data matter to you, keep the original .cr2 — it is the only copy guaranteed to hold the full record. Check the output for yourself before you delete anything.
Yes. Queue as many .cr2 files as you like and they convert with the same Quality Preset and resolution settings, which is usually what you want for a consistent gallery. The practical limit is upload time rather than the conversion: raw files run tens of megabytes each, so a large batch spends most of its wall-clock time moving bytes over your connection.
No. CR3 is a different format despite the similar name: Canon moved to it in 2018 and it uses the ISO base media file format with a custom "crx" codec rather than the TIFF-plus-lossless-JPEG structure of CR2. Files from Canon bodies of the mid-2000s to 2018 era are .cr2 and belong here. For everything a CR2 can become, see the CR2 converter.
That depends far more on the picture than on the camera — fine foliage and noise cost bits, flat skies do not. Two levers matter here: drop Quality Preset from "Very High" to "High" or "Medium", and reduce Image resolution, which usually saves more than quality alone because it removes pixels rather than detail per pixel. If you want to squeeze an existing file further without re-rendering the raw, run it through AVIF compression.