Thursday, August 27, 2020

ISO testing of the Canon R5 and R6 (and comparing them to the Canon 1D X MKIII and 5D MKIV)

Whenever Canon comes out with new cameras, one of my most important real-world tests is determining how clean the images look at higher ISOs. I am not testing this for scientific reasons, I am doing this test because I shoot in low light quite often and want the highest quality images for my clients. I also thought that you and the rest of the world might be interested in this as well.

Many people get caught up in the number of megapixels that a camera has on its sensor, thinking that the more the better. What people may not know is that the more megapixels they cram onto a sensor, and the closer that those pixels are to each other, the more heat build-up occurs. This increase in heat can ultimately also increase the digital noise (graininess) in our photos. 

If we all took photos exclusively outdoors and in bright light, this would not be a concern. But in normal times (not during this pandemic), I spend my weekends photographing in dark temples and churches and even darker party venues. This means that my cameras are usually set at a minimum of ISO 2000 and I sometimes have to raise the ISO as high as ISO 10,000.

For this reason, I was anxious to test the new Canon R5 and Canon R6 mirrorless cameras to see how they perform in low light. And to give me even more insight, I decided to put them to the test against the reigning top Canon DSLR bodies, the Canon 1D X MKIII and the Canon 5D MKIV.

In order to try and keep everything as consistent as possible, I put together a plan to use as much of the same hardware as possible. Since I had the RF adaptor in my possession, I decided to use the Canon 70-200mm 2.8 L lens which could then be mounted to my DSLR cameras and the new mirrorless cameras (using the adaptor). I mounted the Canon 70-200mm lens onto my Gitzo tripod, zoomed it to 200mm and pointed the lens at a dark object in my office (which turned out to be a black NFL helmet in my collection). I decided to set each camera to aperture priority mode at f/8 and vary the ISO levels from 100, 800, 1600, 3200, 6400 and 12,800. I chose these ISOs since they most represent what I typically use for my work. I went as high as 12,800 even though I almost never shoot anything above ISO 10,000.

I printed small cards to put in each shot to make it easier to see which camera and ISO level I was shooting in that particular photo. One by one, I would mount each camera, work my way through the different ISO levels and then switch to the next camera body.

Once done with the entire procedure, I came across some really weird results. But before we get to that, let's look at the noise levels for each camera.

I zoomed into each image to approximately the same spot (varying for the different resolutions). Not surprising, each camera delivered great results at ISO 100, with excellent clarity and very little digital noise. 

Canon R5 at ISO 100  (Image lightened in Photoshop)

Canon 1DX MKIII at ISO 100

The good news is that, even at the higher ISOs (like 6400 and 12,800), the results were very good. Regardless of the camera used, the resulting images from those high ISOs would clearly be good enough for most clients.

Canon R6 at ISO 12,800

If we zoom in and look at the photos, we can clearly see that we are losing detail in the image, but unless these images were printed very big, it would be hard to see this. 

But when we zoom in, this is where we see the biggest difference in ISO performance between not only the Canon R5 and Canon R6, but also the Canon 1D X MKIII and Canon 5D MKIV

Canon 1DX MKIII at ISO 12,800

Canon 5D MKIV at ISO 12,800

Canon R5 at ISO 12,800 (Image lightened in Photoshop)

Canon R6 at ISO 12,800 (Image lightened in Photoshop)

Amazingly, the Canon R6 outperformed all the other cameras, and in a very noticeable way. The R6 image not only retained much more detail than the other three cameras, but also showed much less noise in the lighter and darker areas.

Here is a closer look at the digital noise for the R5 and R6:

Canon R5 at ISO 12,800

Canon R6 at ISO 12,800

I figured that the Canon R6 would outperform the Canon R5 in this ISO test, but was not expecting this $2500 camera to best the $6500 Canon 1D X MKIII by such a wide margin. And here is another thing that puzzles me: If the Canon R6 uses the same sensor and same processor as the Canon 1D X MKIII, why are the ISO results so much better in the Canon R6? I have written to my contacts at Canon to get an answer to this. Stay tuned...

I think that the Canon R6 might be my choice for shooting weddings and bar mitzvah. I do not need 45 Megapixels for my clients and I prefer to have 20 Megapixels and the cleanest image possible.  I also know that, with the in-camera stabilization, I may not need to crank up the ISO quite as high with the Canon R5 and Canon R6

I know that, with the smaller images size that you see here in the blog post, you may not be able to see the details I am talking about. For this reason, I created a Dropbox folder for you all, so that you can download the full-res RAW images to see them for yourself.

So...what were the strange results that I mentioned earlier? After shooting all the images and comparing at the metadata, I saw a wide discrepancy in the shutter speeds for each camera. They were all set for the same ISO, same aperture, and pointing at the same subject through the same lens. Why were the shutter speeds so different? Here are the comparisons:

Shooting in AV mode at ISO 100 and an aperture of f/8, here is what I saw for shutter speeds:

Canon 5D MKIV: 3 sec

Canon 1D X MK III: 4 sec

Canon R5: 1.6 sec

Canon R6: 2 sec

Shooting in AV mode at ISO 6400 and an aperture of f/8, here is what I saw for shutter speeds:

Canon 5D MKIV: 1/15 sec

Canon 1D X MK III: 1/15 sec

Canon R5: 1/40 sec

Canon R6: 1/40 sec

I wrote all this down and stared at the results, trying to figure out how the shutter speeds could be vary so much. With the lower ISO, the Canon R5 and Canon R6 had shutter speeds that were nearly twice as fast as the older DSLR cameras! And at a higher ISO, the numbers (now the same for the new mirrorless cameras and the same for the DSLR cameras), the shutter speeds for the mirrorless cameras was still much faster than the DSLR cameras. I know, from previous conversations with the technical people at Canon, that different cameras with different sensors will yield slightly different results. But these showed a lot more variance than I would expect.

It wasn't until I looked more closely at the images that I saw what was happening.

Look at the image above and you can clearly see that the metering for the DSLR cameras (in this case the 1D X) is quite different from the mirrorless cameras (here showing the R5 images). The mirrorless cameras are determining the metering right from the sensor, whereas the DSLR cameras are determining the metering from a separate sensor. The darker images help explain why the shutter speed was faster on the mirrorless cameras. Now, the big question is...which camera was metering the scene more accurately.  I opened the image from the 1D X MKIII in Adobe Photoshop and hit the "Auto" button to see what it would suggest for an optimum histogram, and it wanted to add +.23 of exposure. When doing the same thing to the R5 image, Photoshop wanted to add +.84 of exposure. (It is for this reason that I lightened some of the R5 and R6 images in the examples above.)

That may come across as technical mumbo jumbo, but the fact is this. I am OK with the exposure captured by the mirrorless cameras, since it gives me a slightly faster shutter speed, and therefore a better chance of having a sharp image. I know I can make the adjustments to the exposure later in post production. 

Here is my thought process at this point: I think I would like to shoot events with the Canon R6 and lean towards the Canon R5 for capturing images on my photo tours. The reason for this is simple, when shooting events, I am almost always in dark environments and don't need more than 20MP. When photographing wildlife in Africa, Costa Rica or other locations, we are usually shooting outdoors in bright light, and I would love to have 45MP of data in those images, for those times when I need to crop in to see a far away subject.

As I write this last paragraph, I got an email from B&H Photo that the Canon R6 is now in stock. I placed my order for that camera immediately. If you are going to be buying anything from B&H. please use my link which is HERE. It won't look any different for you, but it does show me as the referring party which is helpful to me.

Subscribe to the Jeff Cable Photography Blog by clicking HERE!
If you are interested in purchasing ANY equipment, please click here to go to B&H Photo, as I get a referral from them if you enter this way. It does not change the cost to you in any way, but it helps me keep this blog up and running.
Check out my upcoming photo tours to amazing places around the world. I have photo tours to Africa, Costa Rica, Europe, Asia, India and more. And Canon will loan you any gear you want for FREE for any of my tours.


Tuesday, August 18, 2020

More real-world testing of the Canon R5 and R6 cameras - Animal and people eye detection

As part of my ongoing real-world testing of the new Canon R5 and Canon R6 cameras, today's blog post is dedicated to the face and eye tracking feature of the cameras. Unless we are trying to shoot with a creative twist, the goal of a professional photographer is to get our subject's eyes in perfect focus every time. My goal was to push the cameras in extreme conditions and see how they fared. In order to show you the results, this blog includes a lot of sample images. 

I set out to test the tracking of both people and animals with both cameras. I purchased an adaptor so that I could mount my longer EF zoom lenses on the R cameras and started my search to find some subject matter to shoot. Since we are in the middle of this pandemic, my options for shooting team sports is almost zero, so I decided to drive to the coast to see if I could photograph surfers in the ocean.

As I approached the edge of the cliffs above the water, I saw seagulls flying in the area. I used my Canon 100-400mm lens switching between the Canon R5 and Canon R6, made sure the cameras were set to animal tracking and started to shoot.

The eye detection worked remarkably well. Even if I saw a bird flying at me at the last minute, I could quickly raise the camera and still get a nice sharp image. This is true even though the bird is not in the center of my frame. Normally, using my Canon 1DX Mark III or Canon 5D Mark IV, I would be in servo focus mode with the center point activated. This meant that I had to keep my lens right on the bird as it flew by, often yielding images that were out of focus. This was not the case in animal detection mode.

Here is a crop of the same image, showing you the detail of the bird.

It was impressive that the cameras could track the bird, even as it flew through some foliage in the foreground.

Here is another example of how well the camera tracked the seagull, even at the edge of my frame. This would never happen with my Canon DSLR cameras, because there are no focus points in this area of the sensor.

Using the Canon R5 and the same Canon 100-400mm lens, I switched to people tracking and followed some of the surfers in the water.

I locked focus on the woman in the middle of the group to see if the camera would hold focus on her, even with all the other possible subjects surrounding her. 

The tracking was dead on.

It was at this point, with the woman surrounded by other surfers and facing away from me, that I wondered if the camera would jump focus to one of the others. But that did not happen and the camera stayed locked on her.

I asked one of our family friends if he would be my model for an hour and he was nice enough to agree to be my subject for the next test scenario.

For the next test, I wanted to use the Canon R6 and Canon 50mm 1.2 RF lens and use the lens at its widest aperture. For those of you who have ever tried shooting a photo at f/1.2, you know that this wide aperture means that it is very hard to achieve focus on your subject. Any movement of the photographer or the subject means that the subject will be very much out of focus. Whenever I shoot at f/1.2, I expect my "take rate" to be a percentage of the overall photos taken. This is especially true when shooting a couple dancing at a wedding and, with this narrow depth of field, they are swaying in and out of focus. 

I locked focus on Ethan's eye and took a couple of test photos. They were tack sharp on his eye, but this was not the test I wanted. Once again, I locked focus on his eye and this time I asked him to move around.

The tracking worked even at the edges of the frame.

Ethan was moving much more than a typical subject (like a couple dancing in my earlier example), but the camera achieved incredible focus in almost every photo. This is amazing!

This last frame was the only one which was not tack sharp. the eye detection perfect? No, but it is very accurate.  Definitely more so than I would be in this same situation.

This next example is one that I would never try if I were shooting a paid job. I asked Ethan to get on a swing and move at a high rate of speed to and from me, all with me shooting at crazy narrow apertf/1.2.

I locked focus one of his eyes and gave him the go-ahead to start swinging.

The focus held for the first grouping of shots...

...and then I lost focus for the next 3 images.

Soon after, the camera locked back in on Ethan's eye and tracked him once again.

Then I lost focus on him once again...

...only to have the camera lock back in. All of this was happened in a fraction of a second. Once again, the face and eye detection is not perfect, but I know that my "take rate" was significantly better than me using servo focus and trying to keep my focus locked point on Ethan's eye as he went back and forth on the swing.

When we were done, I wanted to capture a nice portrait of Ethan that I could send to him and his parents. I had him sit on a nearby ledge and took this photo of him. And then I had an idea. I wanted to know if a complete novice could capture a good portrait at f1.2 using the eye detection of the new camera. I handed the Canon R6 to Ethan and showed him how to use the back-button focus and asked him to take a photo of me. 

Ethan ended taking a bunch of photos me (with me moving around) and almost every one was in perfect focus.

I can guarantee you that this would not be the case without eye detection. With my other Canon cameras, we would have been lucky to have even one photo turn out!

So...what is my conclusion? The truth is I am totally hooked on both of the new cameras, especially because of the new face and eye detection. I used the Canon R5 on Friday to shoot portraits for a client. I relied on the eye detection for every shot, knowing that the camera would do a great job of getting my subject in focus and freeing me to worry about other things like the posing, exposure, foreground and background.  Whether it is a paying client or my dog running around in the backyard, I can definitely get used to this new technology!

I am in the process of testing high ISO shooting and the usability of the electronic viewfinder with both cameras. Stay tuned for those results in future blog posts.

UPDATE: Adobe JUST released the new Camera RAW which allows us to open RAW (CR3) files from the R5 and R6!!

Subscribe to the Jeff Cable Photography Blog by clicking HERE!
If you are interested in purchasing ANY equipment, please click here to go to B&H Photo, as I get a referral from them if you enter this way. It does not change the cost to you in any way, but it helps me keep this blog up and running.
Check out my upcoming photo tours to amazing places around the world. I have photo tours to Africa, Costa Rica, Europe, Asia, India and more. And Canon will loan you any gear you want for FREE for any of my tours.