Canon 5DSR vs Huawei P20Pro Smartphone. I’m somewhat stunned…

This is the new Huawei P20 Pro smartphone, which boasts what the manufacturers claim is the best ever smartphone camera. Actually, a combination of three sensors with individual lenses, a 40 megapixel (yes!) 1/1.7 inch colour sensor (significantly larger than the 1/2.9 sensor in the iPhone X), a 20 megapixel monochrome sensor (both these sensors are approximately equivalent to a full-frame 27mm lens) and a 8 megapixel 80mm equivalent lens for zooming.

It seems that the three cameras work together and the system uses data from a combination of all three to generate the composite image. This helps drive the phones max ISO up to 102,400 – and this is even before we start looking at the algorithmic tricks it uses for long exposure but stable night shots.

So. How does the resolution on this smartphone camera (albeit a good one) compare with a 50 megapixel dedicated FULL FRAME Canon DSLR? Let’s look!

I put the Canon 24mm f/2.8 IS prime lens on the 5DSR as that would be a reasonable comparable to the 27mm equivalent on the phone (when taking into account the difference between 40 and 50 megapixels). Both were set with everything in ‘Auto’ mode (P mode on the Canon.)



So, here are three pictures:

Top image is the 5DSR, the middle is the P20Pro JPEG, and lowest is the P20Pro DNG file straight out of the camera.  Clearly the P20Pro has some vignetting issues to deal with, but otherwise, not bad.  Also note how it is clearly so much easier to get things straight on a cellphone than with a heavy DSLR!


Of course, at this low resolution even photos taken with a 2000 vintage point-and-shoot are going to look OK, “Show me the crops!” I hear you shout.

Sure thing…  1:1 pixel crops coming up.

Did you guess right?  The upper image is the 5DSR, the lower image is the smartphone.

The crops are taken from the center of the image, as you move out on both systems the quality deteriorates.

Now, there are some things that are clearly obvious such as some purple colour fringes (see the cable running from the security lighting on the right of the lower photo – but you have to agree that the P20 Pro is doing a mighty good job so far.


Next set, this time the order is 5DSR, P20Pro DNG, P20Pro JPG

So, you can see the P20Pro does a quite dramatic JPEG recovery in-camera, fixing the vignetting and boosting the colours and levels to create a rather nice photo (if you want a simple image with no extra work as most phone users do.)

Let’s check those all-important 1:1 crops. I’ve taken the liberty of adjusting the levels on both RAW images first in Lightroom. 5DSR first, P20 Pro following.



Now let’s concentrate on crops. Here’s the original image from the P20 Pro (with numberplate masked out)


And here are the crops from the 5DSR version first, then the P20Pro version

So, color fringing is quite noticeable. This is likely because of the ‘Quad Bayer Filter’.  What this means is that the bayer filter has a resolution of 10 megapixels rather than 40 megapixels, so each block of 4 pixels shares the same colour. This does allow for some good in-camera auto HDR tricks but it does mean when shooting 40 megapixel images you get this fringing.

But let’s disregard that for a moment and instead be astounded at the level of detail that the image has compared to the very much more expensive 5DSR.

And don’t forget, when shooting mono on the P20Pro it switches to the mono sensor (I confirmed this, it’s using the different lens and isn’t simply discarding colour data from the main sensor) which of course has no bayer filter to get in the way.

And, of course, we still have to wait for profiles in Lightroom to improve the P20 Pro raw images. In the meantime, i tried manually. And this is what I got:  I’ve put the 5DSR image again below so you can compare. This time, the P20 Pro photo is at the TOP.


So. In a word. I’m astounded.  Ok that’s two words, but I’m hoping you’ll let me off in these circumstances.

Now, let’s try low-light!

I’m going to try every trick I can, except using a tripod because that’s cheating, to get the 5DSR, which isn’t know for its low light performance, to have a fighting chance. Remember, it’s a $3000+ full-frame camera fighting a cellphone.

So here’s the P20Pro (unaltered) trying to take a photo in our bathroom with the light off and only some reflected light coming through the open door. This was the darkest corner.

And the 5DSR?

Well, after shooting 13 different photos manual focus, I finally got one good photo that could be boosted in Lightroom to give this:

But it was far from easy. The viewfinder was pitch black, focusing didn’t work.

Now, let’s try the P20 Pro’s fabled low light mode. This takes a six second exposure and uses digital stabilization to combine these into a single image. But it does it quite cleverly so that if you have movement in the image it takes a single frame as the base and uses the other images to intelligently update the image.

Now, this does have some artifacts, here’s a 1:1 crop from above.

The artifacts are very similar to those seen when we use the digital zoom (see below), but the end result, when not viewed too closely, is truly impressive.

Zoom is another thing altogether. One annoying thing is that you can’t simply select a particular lens on the P20 Pro and shoot with it in RAW. You only get raw files out of the 40 megapixel colour sensor or, if you switch to mono, from the 20 megapixel mono sensor.

To enable zoom (and to access the 8 megapixel zoom camera) you need to switch back to 10 megapixel resolution, and as you zoom it switches between cropping from the 40 megapixel image to using the 8 megapixel camera as needed.

Unfortunately most people are going to do what I did yesterday and that’s just bang it on maximum zoom and try our luck.

We’re not very lucky it seems.

Here’s a terrible digital zoom at 10x zoom (max) on the phone. I’d suggest sticking to 3x zoom maximum to avoid any digital zooming shenanigans.

So, the conclusion? I certainly won’t be giving up my 5DSR in a hurry, but the quality of the photos from the P20 Pro is astounding. And what’s more, this is a clear indication of the importance of computational photography and how this will become the leading factor in the success of mirrorless cameras as these develop.

Exciting times!

Note. The P20 Pro was bought by me personally and is not a review item or freebie in any way!





You may also like...

24 Responses

  1. Lynn says:

    Hi Jolyon,

    Interesting comparison, thanks! Most people use cell phones for taking pictures of people, and I’d appreciate if your comparison could include a number of portraits/face shots, preferably with people from a variety of races/skin tones. Canon’s colour science is very mature.. how does the Huawei P20 Pro compare?

    Kind regards,

    • jolyon says:

      Simple answer to this is – no, it’s nowhere near as good at portraits as a traditional DSLR. I’d like to do a portrait shoot-out test with the two devices at some point in the future.

  2. Matthew Simpson says:

    Comparisons like this continue to show how unforgivably lax DSLR makers have been when it comes to innovation. Or it could show how much potential the radically larger sensors and better glass could have, if DSLR makers get off their asses before they implode.

  3. J.M. Hoffman says:

    Interesting. Thanks!

    I’m not quite as “astounded” as you are. A year ago I did a similar series of tests comparing consumer-grade cameras to pro equipment, and the results (here: Can You Tell the Difference? Travel Cameras) point in the same direction.

    Purely from the perspective of image quality, there’s less to be gained from pro equipment these days than some people would like to believe…

  4. Alexander says:

    What this means is that the bayer filter has a resolution of 10 megapixels rather than 40 megapixels.

    Hello. Does the Huawei P20 Pro have a real 40 megapixels sensor?
    I am confused.

  5. Jarno says:

    Just curious, what lens did you use on Canon?

    Computational imaging is the future. That’s for sure and DSLRs are the dinosaurs here.

  6. Pat Le Cat says:

    Can you please check if your phone covers the Camera2 interface of Android fully or only partially? There are some probe apps to check for that. Sadly that information is never easy to find if at all. But some pro apps need it.

    • jolyon says:

      Sorry for taking so long to reply:

      Hardware Level Support: Limited

      Exposure: All except “Auto exposure, auto flash redeye” are checked.

      Focus: All checked except “Auto focus EDOF”

      Whitebalance: All checked except “WB: shade, WB: twilight and WB:warm fluorescent”

      Raw capture NOT available.

  7. Hi Jolyon,

    Certainly the smartphone is an amazing camera for what it is, but It seems like at times you’re trying to prove that it is, rather than seeing what’s really happening. The third example, while you call out the fringing, you’ve ignored the mess the P20 is making of the foreground foliage, and the sign on the garage. (the aliasing this camera produces makes all of the images appear sharper, but the detail is false detail). These differences aren’t subtle, and wouldn’t be seen in a DSLR with half the resolution of the 5dsr. I realize this is a blog, and that the “lead” is that the p20 is “surprisingly good”, but….you owe it to yourself and your readers to make good honest observations as part of wanting us to read what you’ve written. The same goes for your lack of any comment about the p20’s inability to render the dark scene, instead switching to the 5dsr’s inability to focus in the dark, and then showing us the p20’s parlor trick, which of course would have little benefit for taking a selfie on a dark street at night.

    Regardless, thanks for introducing us to the P20. It’s clear, that in good light, it’s capable of producing very high quality stills, for a phone.

    • jolyon says:

      You are totally right, and the images speak for themselves. I don’t ever expect a cellphone camera to match up with any DSLR and prime lens in quality, but what I was trying to show was that the 40 megapixel resolution claimed by the P20 Pro was far from a gimmick and is actually useable for real-world stuff.

  8. sim says:

    The thing about this 40mp sensor is that a lot of the time it doesn’t yield nearly as much detail as you’ve shown in the first sample, not even close to the second sample which is already softer. If you check out Anandtech’s samples and GSMArena’s photo comparison tool(which only has a 40mp sample), the smearing is shocking even at something like ISO80, and since the comparison tool is so handy you could compare the results to S9’s and Pixel2XL’s, at 12mp, they both beat the 40mp P20Pro handily under the exact same lighting near base ISO, so maybe the Quad-Bayer eats more color than expected and could only yield that much detail not only with enough light for base ISO but also with a good spectrum(needs clear skies and strong sunshine), otherwise it just smears and smears.
    As for the night mode, your sample is the least smearing I’ve seen anywhere, again Anandtech’s night mode sample smears like pastel, and where there are moving subjects they’re not edited out, there are ghosts of moving people, perhaps you have exceedingly steady hands.

    • jolyon says:

      I don’t have any way of comparing the P20 Pro directly with other phones.

      As I mentioned in the article, I took a bunch of photos in night mode and kept the best, so it’s not what you’re going to get every time you use the feature.

  9. f says:

    You have russian shampoo?

  10. Kurt Härting says:

    What the hell is going on here? The first two crops (brick walls) are taken from totally different angles and at considerably different daytime (just have a look at the shadows!)!
    So at least one of the crops definitely is NOT taken from the uncropped pics displayed at the start as they all share the same perspective and daytime!

    • jolyon says:

      The shots were taken within ten seconds of each other, so that’s not the reason. The difference is due to the perspective difference between using a 24mm lens on the 50 megapixel 5DSR and using a 27mm equivalent lens on the 40 megapixel P20 Pro. The 24mm lens was chosen because it gave approximate the same resolution when cropped in, but there obviously is differences in how the full image is portrayed.

  11. Arup Roy Chowdhury says:

    Honestly I am stunned as well as I also use a A7RII and a Sigma DP3M but the P20Pro’s results are astonishingly good and perfect for social postings that its generally used for. Also it makes photography easy and fun.

  12. Sebastian Müller says:

    Im sorry, but… are you blind? The four photos with the car, have you looked at the foliage? It looks just awful on the P20 like with every “AI” based smartphone camera. More of a psychedelic painting than a photo. Like a lot of people you don’t seem to understand, that most of the picture “quality” comes from heavy post-processing. If it works it looks good, but a lot of times the image gets butchered. Just look at the walls, where the pp has overdone the lines between the bricks.

    In a few years we wont be doing photos at all. We just tell the “camera” what we want, and the app renders something to our likings. God forbid the image shows something real!

    • jolyon says:

      I don’t think I ever intended to suggest the P20 Pro was as good as, or better, than the 5DSR. I mean, I haven’t sold my 5DSR yet 🙂 But compared to other cellphone images, it’s quite astonishing the level of detail that can be seen.

  13. Dan says:

    CA on the phone is horrible. I much prefer the Canon images. No comparison for serious amateur or semi-pro shooter

  14. Returned, quality was not good at all. I bought because i thought that its just a fisheye it doesnt need to be expensive or fancy. well it should at least be decent which this isnt.
    Jerome Nordfalcon

  1. 21st April 2018

    […] Canon 5DSR vs Huawei P20Pro Smartphone. I’m somewhat stunned… – Every Other Shot […]

Leave a Reply

Your email address will not be published. Required fields are marked *