Canon EOS C400 Lab Test – Rolling Shutter, Dynamic Range and Exposure Latitude

Lab Test

by Gunther MachuYesterday

We finally received the new Canon EOS C400 with production firmware (Version 1.0.0.1) at our CineD headquarters. Time to put this promising new contender through strenuous testing in our standardized Lab Test! Curious to see the results? Then keep on reading our Lab Test for the Canon EOS C400!

A lot of us have been caught by surprise by Canon’s announcement of the EOS C80 and its larger sibling, the EOS C400 – especially when looking at the very attractive pricing strategy. A fast full frame 6K sensor with internal RAW recording, high framerates, and Canon’s color science definitely makes for a very interesting package. In case you missed it, read about all the specs for the EOS C80 here, and for the EOS C400 here.

“Redefining versatility, the Canon EOS C400 is a workhorse across cinema, live broadcast, and virtual production.”

– Statement on Canon’s website

As you can see, the footage Johnnie shot in his mini-documentary when he was doing a review on EOS C80, is beautiful. No wonder, the duo was also chosen as two of the “CineD Camera of the Year 2024“.

What I also found on Canon’s website is their claim of dynamic range. It says: “Unleash your creativity with a mighty 6K full-frame BSI sensor and 16 stops of dynamic range.”

Hang on, 16 stops of dynamic range? Time to run this camera through our standardized lab test procedures! As always, a big “Thank you” goes to my colleague Florian, who was a big help in shooting and analyzing this test!

Rolling Shutter of the Canon EOS C400

As per usual, we use our 300Hz strobe light to generate the sequence of black and white bars – a consequence of the read-out nature of CMOS sensors.

Using Canon CRAW at 6K 25 frames per second, we get 9.5ms (less is better) for a 17:9 full-frame image:

That is a very good result for a full-frame sensor – exactly the same that we got for the Canon EOS R3. This value drops to 7.4ms if 4K XF-AVC 120fps is used. In S35 4K crop mode, we get 6.9ms. Please head over to our CineD lab database to see all the results of the different modes.

Dynamic Range of the Canon EOS C400

If you do not know how we test dynamic range, then please head over here first.

In DaVinci Resolve (19, build 69), the Canon RAW files can easily be developed to Canon Log 2 (LOG, CLOG3, and Rec709 are the other options) in the RAW camera tab:

Now, starting with 12-bit Cinema RAW in 6K at ISO800 developed to CLOG2, here is the waveform plot of our Xyla21 chart:

Around 12 stops can be identified above the noise floor, a 13th, and a faint 14th inside the noise floor. In general, CRAW is very noisy. Let’s have a look at IMATEST:

We get 10.5 stops at a signal-to-noise ratio (SNR) of 2, and 11.6 stops at SNR = 1. These numbers and the waveform plot hint at minimal (if any) noise reduction in-camera. In the middle graph above, above the blue “11.6” line, around 3 additional stops can be identified. Also, the “Noise spectrum” on the lower right-hand side shows rather high signal amplitudes at high frequencies – very good!

This camera claims to have a triple base ISO system – interestingly, Canon does not call it “native” ISO – for a good reason as we will see below. In my humble understanding, this suggests that each of the “native” ISO values should be very close in terms of dynamic range – as we have seen previously with a lot of other cameras.

Now, looking at the second base ISO, 3200, we get 9.4 / 10.6 stops at SNR = 2 / 1. Hence, a full stop of dynamic range is lost switching to the second base ISO. Now, looking at the third base ISO value, 12800, we get 8.1 / 9.6 stops at SNR = 2 / 1. Hence, another full stop is lost.

To me, this looks like Canon has optimized these 3 base ISO’s in terms of noise performance, but the sensor doesn’t seem to have the “native” ISO feature.

Now, switching the camera to 4K XF-AVC mode at ISO 800, we get the following waveform plot:

This looks much, much cleaner – obviously, at the expense of in-camera noise reduction. Let’s have a look at what IMATEST calculates:

We are getting 13 / 14.1 stops at SNR = 2 / 1. Hence, a very good out-of-the-box result that can be used without postprocessing!

You can see very nicely in the “Noise spectrum” graph above, that fine details are lost at higher frequencies (amplitudes drop to ~0.3 at 0.3 frequency versus CRAW where at the same frequency still an amplitude of ~0.5 is present).

At ISO3200, we get 11.4 / 12.8 stops at SNR = 2 / 1, and at ISO 12800 we get 10.8 / 12.1 at SNR = 2 / 1.

Exposure latitude of the Canon EOS C400 at ISO800

As said before, latitude is the capability of a camera to retain detail and colors when over- or underexposed and pushed back to a base exposure. This test is very revealing, as it pushes every camera to its absolute limits – not just in the highlights but also in the shadows.

All latitude shots were done at ISO800 with 6K 12bit Canon RAW developed to CLog2 in the Camera Raw tab in DaVinci Resolve.

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

Now, let’s jump to 4 stops of overexposure:

As can be seen in the 4 stops overexposed shot, the red channel on Johnnie’s face is at the cusp of clipping but intact.

In the Camera RAW tab in DaVinci Resolve, it’s very easy to push the files back to base exposure. Just use the exposure slider and adjust. Unfortunately, though, this only works from +3 to -3. Above or below that, the typical lift, gamma, and gain controls were used. Also, a 3-node tree was used with the first node converting the CLog 2 files to DaVinci wide gamut (via a color space transform), then in the middle node all adjustments are made, then in the final node, another CST is used to get to Rec709. Noise reduction was always made on the first node.

Now, let’s start to underexpose by closing down the iris of the lens in one-stop increments (until T8-after that the shutter value was halved).

At 3 stops under and pushed back, we see significant noise creeping into the image:

Also, some horizontal lines appear. Not very distracting yet, but definitely there.

At 4 stops under, pushed back we get the following image:

The noise becomes dominant, and without noise reduction, this image is not usable anymore. We are at 8 stops of exposure latitude, typical for consumer full frame cameras (like the Canon EOS R3 for example).

So let’s apply noise reduction:

It cleans up OKish, but there are these horizontal lines that are hovering around in the moving image which are very distracting. Also, the noise is not very finely dispersed, it appears rather coarse. I would have expected a much finer noise footprint with 6K RAW footage.

Let’s move to 5 stops underexposure, pushed back. We are at 9 stops of exposure latitude now:

Here is the noise-reduced version with the settings in DaVinci Resolve:

I was playing around for quite some time to get the noise more or less under control (using the new UltraNR feature in Resolve 19).

The image gets a pinkish cast, and there are larger horizontal bands of pink and green chroma noise, plus the horizontal lines which are very prominent now, hovering all over the place in the moving image. Although the shadow side of Johnnie’s face is still somewhat okay, these large bands plus the horizontal lines cannot be tolerated anymore. It’s game over!

Just for your reference, here is the 6 stops underexposed image, pushed back to base exposure:

In summary, that gives us 8 stops of exposure latitude with some room towards 9. This was state-of-the-art for consumer full frame cameras some time ago and is very very similar to the Canon EOS R3.

However, more recently some consumer cameras like the Sony A9 III using a global shutter full frame sensor have shown 9 stops of exposure latitude. Not to speak of the RED V-RAPTOR VV and V-RAPTOR [X] VV which also have a solid 9 stops (but are in a different price bracket). Also, the Sony BURANO 8K showed close to 10 stops of exposure latitude.

Hence, the EOS C400 disappoints somewhat in that regard… especially when having Canon’s claim of “16 stops of dynamic range” in mind.

The current leader of the pack for full frame cameras is the ARRI ALEXA Mini LF with 10 stops of exposure latitude, and for S35 sensors the ALEXA 35 exhibited 12 stops of exposure latitude.

Summary

The Canon EOS C400 shows a solid but average result in our Lab Test. It exhibits a very good rolling shutter of 9.5ms but falls behind in the dynamic range department. The Canon RAW files are very noisy, and it is difficult to reduce the noise in postproduction. Using the 10bit XF-AVC compressed codecs, the in-camera noise reduction works very effectively to give ready-to-use files with a very good dynamic range performance, albeit sacrificing some details.

Our real-world studio scene, namely the exposure latitude test again reveals the shortcomings of the image pipeline from sensor to codec: we get a good 8 stops of exposure latitude, but already in 2024 that is not state of the art anymore. In the meanwhile, a lot of other consumer cameras are already topping that. Even the recently tested Panasonic LUMIX GH7 exhibits a rather similar result in the exposure latitude – using a 5.7K Micro Four Thirds sensor, which is 4x smaller than Canon’s EOS C400 full-frame sensor.

Have you shot with the Canon EOS C400 so far? What are your experiences? Please let us know in the comments section below!