Convert CR3 to WebP

It is worth saying plainly, because the usual first assumption is that something went wrong with the card. Nothing did. CR3 is the container Canon moved to with the EOS R generation in 2018, and it is structurally unrelated to the CR2 files older Canon cameras wrote — the file begins with an ISO base media header, the same family of structure as a video file, rather than the TIFF-based layout that made CR2 at least partly legible to other software.

Canon did not publish the specification, so every other program had to work it out, and that took years. The consequence for a school or club laptop is absolute: the built-in photo viewer will not show a thumbnail, the browser will not display the file, and the website will reject the upload. The photographs are perfectly good; the container is the problem, and converting it is the whole fix.

Every EOS R body writes a complete, full-resolution JPEG inside the CR3 at the moment the shutter fires — it is what the rear screen shows and what a laptop would use to draw a thumbnail if it could read the file. This converter finds that picture and re-encodes it as WebP.

So what you get is the camera’s own rendering, with whatever Picture Style, white balance and noise reduction the body was set to. For a club gallery that is exactly right: the pictures look like the ones everyone saw on the back of the camera on the day. What it is not is a development of the raw data, so if a frame is badly underexposed, this route cannot rescue it — that needs Canon’s free Digital Photo Professional and somebody willing to learn it.

The realistic obstacle for most organisations is not the format, it is permission. Installing software on a school or council machine means a request, an approval and a wait, and Canon’s own utility is a several-hundred-megabyte download that somebody has to be persuaded to authorise.

This runs in the browser tab you already have open. There is no installer, no account, no administrator prompt and nothing added to the machine, so it works the same on a locked-down Windows laptop, a Chromebook, a Mac or a tablet. It also works on somebody’s home computer on a Sunday evening, which is when a good deal of small-organisation website updating actually happens.

This is the setting that matters and the one most people skip. An EOS R6 frame is 5,472 pixels wide and an R5 frame is 8,192; a website displays photographs at perhaps 1,000. Uploading the full size means every visitor downloads roughly thirty times more picture than their screen can show, which makes the page slow for them and eats the hosting allowance.

Set a maximum width of around 1,600 during conversion. That is generous enough for a large monitor and a high-density display, and it removes about ninety per cent of the pixels. If the pictures are only ever shown as thumbnails in a grid, 800 is plenty. Do this once at the start and apply it to the whole batch, so the gallery is consistent.

At the same visible quality WebP is typically a quarter to a third smaller than JPEG. On one photograph that is nothing; on a gallery of forty it is a page that loads while somebody waits at a bus stop rather than one they give up on. It is also the single easiest thing a non-technical site editor can do about page speed, because it requires no changes to the site itself.

The format has been published since 2010 and every current browser reads it, so there is no fallback to arrange. The one place to be careful is email: some mail clients still will not render a WebP, so if the same pictures are going into a newsletter, produce JPGs for that and keep the WebP files for the website.

The conversion decodes the picture and encodes it again, and no metadata survives that. The published images carry no capture date, no camera or lens, no serial number and no GPS coordinate.

For an organisation publishing pictures of children, members, patients or service users, that is a genuine and easily overlooked benefit. Camera bodies and phones tag location more often than their owners realise, and a gallery of photographs from a school event that quietly states the school’s coordinates and the exact times each picture was taken is more information than anyone intended to publish. Consent and safeguarding are still policy questions the converter cannot answer; this only removes the accidental part.

The output takes the source name, so IMG_4172.CR3 becomes IMG_4172.webp. On a website the file name is one of the few pieces of text attached to a picture, and a camera counter tells a visitor and a search engine nothing at all.

Rename the set to something descriptive before uploading — the event, the year, a number — and write real alt text for each picture in the site editor. Alt text is also what a screen reader announces, so it is an accessibility obligation for a public body as well as a small search benefit. Neither of these takes long on forty files and both outlast the gallery.

Drop the entire selection at once rather than one file at a time. Each CR3 is handled separately with its own progress, and the finished WebP files can be taken together as a single ZIP, which is much easier to hand to whoever is doing the uploading than four hundred separate downloads.

Input files are accepted up to 100 MB on the free tier and every CR3 clears that comfortably — a 45-megapixel R5 frame is around 45 MB, and about half that if the camera was set to Canon’s compressed raw. The work happens on your own machine, so a large card is limited by the laptop rather than by a queue.

The most useful thing on this page may be a suggestion for the next event. Every Canon body can record a JPEG alongside the raw file, or instead of it, and if nobody in the organisation is going to develop the raw files there is very little reason to be collecting them. Setting the camera to JPEG, or to raw plus JPEG, removes this entire step.

Keep shooting raw if somebody does occasionally rescue a frame, or if the photographs might be printed large later. Otherwise the card fills four times faster, the transfer takes four times longer, and the person updating the website has to convert everything — which is a lot of friction to carry for an option nobody uses.

The CR3 files are read from disk by this page, scanned for the embedded picture, decoded and re-encoded as WebP, all locally. No request carries an image, which you can see in the browser’s network tab or prove by disconnecting from the internet and converting anyway.

That is worth checking rather than taking on trust, because most free converters do upload, and for an organisation handling photographs of identifiable people the difference is not academic. A tool that sends the files to an unnamed server is a processor you have not assessed, and using it is a decision somebody would have to justify.