Canon EOS R5 Mark II Lab Test – Rolling Shutter, Dynamic Range and Exposure Latitude

Lab Test

by Gunther MachuMarch 10^th, 2025

Canon’s successor to their popular EOS R5 model has arrived in our lab, so we took the new EOS R5 Mark II camera through our standard CineD lab test procedures. Curious to hear how the Canon EOS R5 Mark II lab test went? Then read on!

Canon’s EOS R5 Mark II camera packs a punch of features into its tiny body: it offers 8K up to 60fps with 12-bit raw video internally, as well as 4K 120fps with sound, 4-channel audio, IBIS, better integration with Canon Cinema EOS formats, Canon Log 2, a new cooling grip for longer recording times, and more …

Please read more about the full specs here. Quite amazing, actually!

Now it is time to run it through our standard CineD lab test procedures. A big shout-out goes again to my colleague Florian Milz, who helped shoot and analyze the lab tests!

Canon EOS R5 Mark II lab test – rolling shutter

As always, we use our 300Hz strobe light to reveal the sequence of black and white bars, typical for the read-out nature of CMOS sensors.

Let’s have a look at 8K DCI (17:9) first:

We get a rolling shutter of 17.3ms (less is better). That is a tad worse than the first generation EOS R5, which exhibited 15.5ms but was still good.

In full frame 4K “Fine” mode, we get the same 17.3ms result. If we switch to the “normal” 4K mode, rolling shutter drastically improves to 9.7ms:

It seems that 4K “fine” subsamples the full 8K sensor to 4K, versus the “normal” 4K mode which does some sort of binning or line skipping to achieve the faster readout speeds. This result of 9.7ms is also seen in the 4K 60p mode.

So far so good. In 4K 120p, rolling shutter further improves to 7.1ms.

Canon EOS R5 Mark II lab test – dynamic range at ISO800

Let’s start with 8K internal 12bit Canon RAW (CRAW) recording, developed to CLog 2 on an 8K timeline at 25 frames per second in DaVinci Resolve 19:

The waveform plot exhibits 12 stops above the noise floor. Inside the noise floor, we can identify a 13th and a 14th stop. Let’s have a look at IMATEST:

As can be seen, at a signal to noise ratio of 1 (SNR) we get 12 stops, at SNR = 2 10.8 stops. As is typical for Canon RAW, the image is quite noisy, really “RAW” and we can identify about 3 more stops inside the noise floor in the middle graph shown above, above the blue “12” line.

Also, the lower right-hand side graph “noise spectrum” shows rather fine detail. Only towards 0.5 frequency, amplitudes drop to around 0.2.

Now let’s look at internal 10bit 8K XF-HEVC, 25p CLog 2 at ISO800:

We now get 13.1 stops at SNR = 1, and 11.9 stops at SNR = 2. This comes at the expense of internal noise reduction, as can be seen in the “Noise Spectrum” graph: already at a Frequency of 0.2, amplitudes drop below 0.2. A very nice demonstration of the effect of noise reduction: higher frequency details (i.e. higher resolution detail) look like they are washed out.

Now let’s look at 10bit 4K “Fine“:

The waveform now exhibits 13 stops above the noise floor – a clear consequence of the internal noise reduction happening with compressed 10bit modes, effectively digging out these stops from the noise floor.

IMATEST confirms this result:

Now we are getting 14.2 stops at SNR = 1, and 13.3 stops at SNR = 2! Those are really high numbers, the best we have seen for consumer full-frame cameras in this price bracket.

As usual for Canon cameras, if you want the highest detail/resolution, you have to shoot in CRAW. Out-of-the-box, superb dynamic range results can be achieved using the internal compressed 10bit XF-AVC or XF-HEVC modes.

On a separate note, we have seen above that the full frame readout speed improves massively if the “normal” 4K readout mode is used. Interestingly, looking at the dynamic range results, virtually no difference can be seen at SNR = 1 and 2, results are 14 / 13.1 stops. Hence, the “normal” 4K mode looks like it is a great documentary shooting mode: very low rolling shutter with a great dynamic range out of the box. The only caveat could be less resolution / moiré effects due to the line skipping / pixel binning.

For the other modes, have a look at our CineD database.

Canon EOS R5 Mark II lab test – exposure latitude

As stated in previous articles, latitude is the capability of a camera to retain details and colors when over- or underexposed and pushed back to a base exposure. This test is very revealing, as it pushes the complete image pipeline of any camera to its absolute limits – not just in the highlights but mainly in the shadows.

We developed the CRAW files in DaVinci Resolve 19, then using a three node tree with a color space transform (CST) to DaVinci Wide Gamut / Intermediate on the first and another CST Rec709 Gamma 2.4 transform on the last node:

Our studio base exposure is (arbitrarily) chosen as having an (ungraded) luma value of around 60% on the forehead of our subject on the waveform monitor – in this case, my colleague Johnnie:

I really like the way the colors come across. Very natural and perfect skin tones.

From here, we can overexpose by 4 stops before the red channel is at the cusp of clipping (on Johnnie’s forehead), then bring it back in post to the base exposure using the exposure slider in the camera raw tab of DVR (which works from +3 to -3), plus adjusting the ISO:

Now, from the base exposure we stop down by three stops and push the image back to base:

This looks good; we are at 7 stops of underexposure. Noise starts to creep in, and at the lower end of the image some faint horizontal stripes appear.

At 4 stops of underexposure, pushed back to base we get the following image:

We are at 8 stops of exposure latitude and noise is now really affecting the image. Shadows are getting a pinkish/purple cast, and larger blotches of chroma noise can be seen. Also, the horizontal stripes are much more pronounced. Noise reduction will not help with the horizontal stripes, as can be seen below:

Basically, this looks tolerable as the noise cleans up well – I always look at the shadow side of the subject’s face to see if skin tones are still okay or not. But fixed patterns like the horizontal lines are very distracting in a moving image, hence cannot be ignored – looks, like this is the Achilles’ heel of the EOS R5 Mark II and the limiting factor for the exposure latitude test.

For your reference, the 5 stops underexposed, pushed back image looks like this:

And here the same image, using noise reduction:

Using UltraNR in DaVinci, but reducing the luma component massively – otherwise everything becomes very plasticky and details are gone – we can see that the chroma noise cleans up OK’ish but the horizontal patterns are the issue. Not to forget about the purple/pink cast across the whole image.

In summary, we can say that the Canon EOS R5 Mark II exhibits around 8 stops of exposure latitude. That is a typical result nowadays with recent full frame consumer cameras, but we have to mention there are already examples of full frame cameras with 9 stops latitude available (like the Sony A9 III for example, lab test here). Well, even recent Micro Four Thirds cameras like the recently tested Panasonic LUMIX GH7 manage 8 stops of exposure latitude (lab test here).

Summary

The Canon EOS R5 Mark II is a very solid performer in the lab. Rolling shutter values are okay, if not superb (using the “normal” 4K full frame mode), and the dynamic range is solid using the CRAW mode, but requires postproduction as it is very noisy out of the box. Earlier, Canon cameras like the EOS R5 or R5C were even worse, so things have improved. On the compressed codecs side, using the 10bit 4K mode gives great dynamic range values out of the box without the need of much postprocessing.

Looking at the exposure latitude results, in essence, images look great until 8 stops of latitude, however, be aware that the EOS R5 Mark II is hampered by purple shadows and fixed horizontal stripes when pushed to the extreme.

Have you already shot with the Canon EOS R5 Mark II? Are you planning to add this little gem to your arsenal of tools? Let us know in the comments below.