CR3 to PNG

CR3 is the raw format Canon has written since the EOS R and M50, and plenty of software that opens a CR2 still sees a CR3 as an unknown file. This page decodes it in your browser and writes a PNG — the output to choose when the picture is going into an editor, a layout or an archive and should not pick up JPEG compression on the way.

Drop CR3 images here

CR3 files — or any of PNG, JPEG, WebP, AVIF, BMP, ICO, TIFF, GIF or camera raw (CR2, NEF, ARW, DNG, RAF, ORF, RW2). Everything runs on this device.

Or paste an image from the clipboard.

Your images never leave this device. They are decoded and re-encoded by your own browser, and the security policy on every page blocks this site from sending them anywhere.

Convert to

Lossless, keeps transparency, universally supported, larger than modern formats.

Quality

PNG is lossless — it stores every pixel exactly, so there is nothing to trade away.

Size

Images keep their original dimensions.

Rotate

Images are turned upright from their own orientation data, and left that way.

Transparency

Kept. PNG stores a full alpha channel, so cut-outs and soft edges come through exactly.

Metadata

Exif is carried across where both formats can hold it. Nothing currently in the queue has any.

Sensor detail

Decodes at the file's full resolution, interpolating missing colours when it has a colour filter array. Files already holding colour pixels need no demosaicing.

White balance

The value your camera recorded when the shutter fired — what you saw on the back of the camera.

Camera, lens, exposure and capture time are written into the PNG when those fields are present. The file details show whether GPS data exists, but coordinates are deliberately not written, and colour comes out as sRGB.

What happens to your image

File size depends mostly on the image itself.

What is preserved and what is lost converting CR3 to PNG
Property	Outcome
Transparency	Preserved: CR3 images have no transparency to lose.
Image quality	Preserved: Unchanged. Both formats are lossless, so every pixel comes through exactly.
Colour	Partly preserved: LibRaw develops the sensor data into eight-bit sRGB. The output uses that colour space; a source colour profile is not copied into the output.
Exif metadata	Partly preserved: When metadata is switched on, the camera, lens, exposure and capture time recorded in the raw file are assembled into a standard block and written into the PNG. Switch it off to remove them. The GPS position is shown to you in the file details and deliberately never written.
Orientation	Preserved: Applied to the pixels, and the tag is reset to upright afterwards so nothing rotates the image a second time.

Full raw, CRAW, and what lossless can mean

Canon bodies offer two raw qualities. Full raw stores the sensor readings without loss; CRAW is smaller because the camera discards some information before writing the file. Both are decoded here from their sensor data, and nothing can bring back what CRAW left out.

From the decoder onwards the PNG is lossless: every pixel LibRaw produces is stored exactly, and saving the file again loses nothing. What the PNG cannot hold is the raw data itself. It is eight bits per channel in sRGB, with white balance, the colour matrix and the tone curve applied, so keep the CR3 as the archive and treat the PNG as a high-quality working copy.

Why a CR3 needs its own decoder

A CR2 was a TIFF with lossless-JPEG sensor data. A CR3 is an ISO base media container — the family MP4 belongs to — holding sensor data in a compression scheme Canon introduced with it. Support for one says nothing about the other.

Here the file is identified from its own bytes, its metadata read for the true sensor dimensions, camera, lens and exposure, and the memory a full decode needs worked out before anything is allocated. Then the sensor data is unpacked, the Bayer mosaic interpolated with adaptive homogeneity-directed demosaicing, the camera's colour matrix applied and a tone curve mapped. The PNG is reopened and compared with the pixels that went in before it is offered.

The camera also stores a full-size JPEG preview inside a CR3, almost the same size as the sensor image. It is used only if the sensor data genuinely cannot be decoded, and then the row, the result and the ZIP manifest all say the output came from the preview.

Expect a large file

A PNG stores three full channels per pixel and cannot throw away noise or fine detail, so it is usually two or three times the size of the CR3. The EOS R6 test file this page is checked against is a 5 MB CR3 at 3407 × 2271, and it becomes a 14 MB PNG — nearly three times the size. A detailed frame from a 45-megapixel R5 is much larger again.

If the files are going to be viewed rather than edited, the CR3 to JPG page gives a result a fraction of the size that looks the same on a screen. Half resolution is the other option: a quarter of the pixels, a quarter of the memory, and no interpolation step.

If something goes wrong

A file from a very recent Canon body will not decode.

Canon has changed CR3 details between camera generations, and a body newer than this decoder is the one case that cannot work.

What to do: Adobe's free DNG Converter can usually turn it into a DNG that converts here. The message names the camera so you know which files are affected.

The PNG looks flatter than the camera's own JPEG.

The camera applied its Picture Style, sharpening and noise reduction. Only the camera matrix and a standard tone curve are applied here.

What to do: Expected. Apply the look in your editor; the PNG holds every decoded pixel, with no compression artefacts in the way.

A Dual Pixel raw file converts to only one image.

A Dual Pixel CR3 holds two slightly different views from the two halves of each photosite. Only the primary image is decoded.

What to do: Expected. The second view is only useful to Canon's own software, which uses it for Dual Pixel adjustments.

Questions

Why PNG rather than TIFF?

PNG opens in practically everything, including every browser, and is losslessly compressed by default. This tool does not write TIFF. If you specifically need sixteen-bit TIFF for heavy editing, a raw editor is the right tool for that step.

Are the camera and lens details kept?

With metadata on, the camera, lens, exposure and capture time read from the CR3 are written into the PNG's Exif chunk. GPS coordinates are shown in the file details and never written. Some older programs ignore Exif in PNG.

Does this work on an iPad or iPhone?

Yes, in Safari. Raw decoding is memory-hungry, so files convert one at a time there and very large ones are decoded at half resolution, with the row saying so.