Canon EOS R5 Lab Test - To Film in RAW or not?
Cover Image.jpg

When I looked at the specifications of the Canon EOS R5, my first reaction was, "Who needs 8K RAW?" – as this camera also offers high quality 4K modes. Until I see the lab results … Curious to find out more about our Canon EOS lab test? Continue reading.

At the beginning of August the Canon EOS R5 arrived for a short visit to the CineD office and showed the firmware version 1.0. I was really curious to see how this mirrorless camera would perform in the lab as its specs are amazing despite the overheating problem reported early on.

Rolling shutter, dynamic range and our underexposure test were carried out – which surprisingly turned out to be a challenge as the overheating warning popped up at the beginning of the test setup – and some test runs had to be repeated after a waiting period of 15 minutes as the camera continued to shut down between shots.

That fact diminished my enthusiasm for the camera significantly – but that's another story.

If you want to know how we test dynamic range, have a look here.

Canon EOS R5 – roller shutter

Interestingly, despite the high resolution of the Canon EOS R5 sensor, a very good readout performance for roller shutters was recorded.

The Canon EOS R5 in full-screen DCI mode (17: 9) in high-quality readout mode (ie 8K RAW, 8K H265 All-I and 4K DCI) has a rolling shutter of 15.5 ms. This is the best result that we have measured so far for a full-frame sensor, even slightly better than the Canon C500 Mark II (15.8 ms).

Fig. 1: Rolling shutter measurement for the Canon EOS R5 in 8K 25fps: very good 15.5 ms are measured.

Rolling shutter for full screen 25fps low quality and 50 frames per second (in 4K DCI) were the same 9.6 msHence, there is obviously undersampling in the background.

Canon EOS R5 – dynamic range at ISO 400

As usual, I started testing the camera's internal H265 modes, but I've found that 8K RAW looks different and less high-contrast (despite using Canon LOG for H265).

So I tested both 8K RAW and the different H265 modes (8K DCI and 4K DCI). And yes, the distribution of the code values ​​and the results are different – in 12-bit 8K RAW, the clipping luma value is 80%, while 10-bit H265 is 100%.

In addition, I was unable to use the 8K files directly in IMATEST, as I first had to develop the files in DaVinci Resolve (16.2.5), export TIFFs and then do the analysis. Unfortunately, my GTX980 graphics card with 8 GB of memory showed a "GPU error" in the native resolution. So I had to downsize to 6K for it to work.

Unsurprisingly, if you want to work with the native resolution 8K RAW files, you will need very powerful hardware.

Let's start with 8K RAW. Here is the waveform diagram (Figure 3). As can be seen below, about 12 stops can be identified above the noise floor. In general, the 8K RAW files are VERY noisy. Nevertheless, a 13th and even a 14th stop is visible in the background noise.

Fig. 3: Waveform diagram of the Xyla21 stepchart for the Canon EOS R5 at 8K RAW, ISO 400 25fps. About 12 stops can be seen above the background noise.

Because 8K RAW is so noisy, IMATEST (Figure 4) calculates the signal-to-noise threshold (SNR) of 2 at 10.3 stops.

Fig. 4: IMATEST dynamic range result for the Canon R5 at 8K RAW ISO400 (scaled to 6K): 10.3 stops are calculated for SNR = 2, but more stops are buried in the noise floor (red circle in the middle graphic) with some noise suppression can be retrieved in the mail.

However, a patch area of ​​16.8 stops is identified, and beyond the SNR = 1 (blue curve; "Low") in the middle graph (Figure 4) on the left side at least 3 to 4 additional stops (buried in the Background noise) are present. These can be "dug up" in the post with a certain amount of noise reduction. So despite the low 10.3 stops at SNR = 2, there is a lot more if you know how to handle the files in the mail.

Obviously the RAW files really represent some RAW outside of the sensor – there doesn't seem to be a lot of signal processing (noise filtering etc) going on. Which is a good thing!

Big surprise When testing the H265 8K mode at ISO400 it looks like the noise floor is almost cut off and only 10 stops are calculated by IMATEST.

Fig. 5: Waveform diagram of the Xyla21 step diagram for the Canon EOS R5 at 8K H265 ISO400 CLOG: About 11 f-stops are visible above the noise floor, and an indication of a 12th f-stop. The distribution of code values ​​is fundamentally different from that in RAW mode, and more than 12 stops are inaccessible.

We have to speculate what is happening here and why the darker stops are no longer accessible. If we look again at the middle graph of the IMATEST dynamic range results, this behavior is clearly visible (Figure 6) – where I mentioned the red circle in Figure 4 above, in Figure 6 the noise signal is gone.

Fig. 6: IMATEST result of the Canon EOS R5 dynamic range in 8K DCI H265 ISO400 CLOG (scaled to 6K): 10 stops are calculated for SNR = 2, and the identified total area of ​​the patch is 11.8 stops. The remaining stops visible for RAW (Figure 4) have been cut off.

A similar, if slightly better result is obtained for the 4k DCI mode (high quality, full screen). 10.8 stops are calculated at SNR = 2, see Figure 7 below (probably due to the scaling of the 8K in 4K).

Fig. 7: IMATEST result of the Canon EOS R5 dynamic range in 4K DCI H265 ISO400 CLOG: For SNR = 2 10.8 stops are calculated, and the identified total patch area is 12.8 stops. The remaining stops visible for RAW (Figure 4) have been cut off.

The dynamic range results achieved for the Canon EOS R5 therefore show a mixed bag: In 8K RAW, if you have the time for the, you can get further stops beyond the 10.3 steps at SNR = 2, albeit very loud Want to take post production. If you want to use H265 in 4K DCI mode, you'll get a slightly better result at 10.8 stops (SNR = 2), but you're sacrificing everything beyond stop 12.

Next up is our underexposure test. As already mentioned, we are arranging a studio scene in which one of my colleagues appears as a model – in this case Nino – and his face at max. 60% luma value for a shutter angle of 360 ° (or 1/25 of a second) at F4 for 25 fps 8K RAW ISO400.

As I noticed in the dynamic range tests on the Canon EOS R5, the 8K RAW files are flatter and have more stops. So I was curious to do this test in 8K RAW.

Then we underexpose the scene one after the other by reducing the shutter angle to 180 °, 90 °, 45 °, 22.5 ° and 11.25 °, thus achieving 5 underexposure stops.

Post-production with DaVinci Resolve 16.2.5 then brings the files back to the zero exposure baseline. This is quite convenient with RAW, as only the "Exposure" slider on the RAW tab of Resolve needs to be adjusted. Oddly enough, this only worked up to +3 stops. The slider allowed +4 and +5, but nothing happened. So I had to manually adjust the image for +4 and +5 by raising, gamma and gain controls (checking the waveform of the respective scene as a reference).

It's also important to note that I ran these underexposure tests on a 2K DCI timeline in Resolve (otherwise my 980 GTX GPU gave errors using noise cancellation). As a result, the 8K RAW files are already scaled to 2K, which massively improves the noise using DaVinci Resolve's advanced scaling algorithm alone.

So here is the zero baseline scene (slightly graded):

Fig. 8: The zero baseline scene for the Canon EOS R5 at 360 ° shutter, F4, 8K RAW ISO400.

At the underexposed scene, let's jump forward 3 stops and slide back to zero (Figure 9). The image can still be used, but there is already a lot of noise. The noise reduction helps to improve the picture, see Figure 10 (3 frames in time, threshold 30 and spatial threshold 8).

Fig. 9: The Canon EOS R5 at 8K RAW ISO400, 3 f-stops underexposed and pushed back to zero. There is a lot of ugly noise despite the reduction to 2K – as already mentioned, 8K RAW is very loud.Fig. 10: The Canon EOS R5 at 8K RAW ISO400, 3 f-stops underexposed and pushed back to zero, plus noise reduction in Resolve. You can get very usable results.

With 4 underexposure stops, the image falls apart due to the strong noise (Figure 11). In addition, the noise reduction does not help because it has to be so strong that the image details suffer massively.

Fig. 11: The Canon EOS R5 at 8K RAW ISO400, 4 f-stops underexposed and pushed back to zero. The noise is so strong that I found it almost impossible to remove it by noise reduction without affecting the image details.

For reference only, here is the 5-stop underexposed image that has been pushed back to zero (Figure 12). Color and detail are quite well preserved, but the noise …

Fig. 12: The Canon EOS R5 at 8K RAW ISO400, 5 f-stops underexposed and pushed back to zero.

Also, just for fun, I tried a strong temporal (4 images, threshold = 35) and spatial (threshold = 10) noise reduction for the underexposed image with 5 f-stops, see Figure 13. The image details have completely disappeared.

Fig. 13: The Canon EOS R5 at 8K RAW ISO400, 5 f-stops underexposed and pushed back to zero. The noise is so strong that I found it impossible to remove it using noise reduction without greatly affecting the image details.

In summary, I would rate that -3 stops underexposure Picture as that usable limit.

The laboratory results show a mixed bag: The performance of the rolling shutter on the EOS R5 (15.5 ms) is outstanding despite the high resolution – the best we have measured for a full-frame camera so far!

The dynamic range of the Canon EOS R5 in the CLOG modes H265 is at the lower end for a full-frame sensor – only in 8K RAW does the Canon EOS R5 start to shine by offering additional stops, albeit buried in the background noise.

In the underexposure tests of the Canon EOS R5, the performance is average even though we used 8K RAW. 3 Aperture underexposure is possible, but not more. Even smaller sensor cameras offer more in this department (e.g. the BMPCC6K). Please see our section on laboratory tests for more information.

8K RAW is very, very stressful on computer hardware. Unless you're using the latest hardware, you'll run into massive problems during playback, and not even mention of extra nodes for color correction.

Also, overheating was a big problem during the lab test: just setting up the measurements and the studio scene, keeping the camera running and recording videos was very problematic.

So choose wisely if this camera is right for you – especially given the various other offerings on the market.

The Canon EOS R5 is already being shipped. Do you have any experience with filming? Please tell us some stories from the archive!

LEAVE A REPLY

Please enter your comment!
Please enter your name here