Canon C500 Mark II Lab Test: Dynamic Range, Latitude and Rolling Shutter

April 15th, 2020
Canon C500 Mark II Lab Test: Dynamic Range, Latitude and Rolling Shutter

Having tested now various mirrorless full-frame cameras for the last 2 years in our cinema5D lab, I was really curious to put a “real” cinema camera with a full-frame sensor through the paces one day. This day had finally arrived when the Canon C500 Mark II was available in the cinema5D headquarters earlier this year. Will it take the crown in our lab tests? Read on …

A lot has been said and done on the Canon C500 Mark II already, for example by my colleagues Nino or Ollie here or here.

So without further delay lets jump right into the lab test …

Dynamic Range Result of the Canon C500 Mark II at ISO 800

Please have a look at how we test dynamic range of cameras here.

The dynamic range of the C500 Mark II using the internal 10bit XF-AVC codec at UHD 25fps using Canon Log2 / C.Gamut (Color Matrix “Neutral”) shows a very strong 13.1 stops at a signal to noise ratio of 2, see figure 1 below (noise reduction “off”). This result comes directly behind the ARRI Alexa (14 stops), and is the second-highest value we measured so far – pushing the Panasonic S1H (12.7 stops) from the second to third place!

Fig. 1: IMATEST dynamic range result for the Canon C500 Mark II at 25fps, UHD, 10bit XF-AVC codec (CLOG2 / C.Gamut) at ISO 800

The waveform of the Xyla21 stepchart confirms this:

Fig. 2: Waveform plot of the Xyla21 stepchart from the Canon C500 Mark II at 25fps, UHD, 10bit XF-AVC codec (CLOG2 / C.Gamut) at ISO 800 – about 13 stops can be identified above the noise floor!

Now you may ask why I have not used the internal 6K RAW (5952×3140) to test the dynamic range? Well, I did. And got the same result as with other recent raw cameras before: using plain raw without any postprocessing, the images are very noisy. Hence, the dynamic range results from IMATEST are lower – indicating a rather “untouched” image by any in-camera processing.

Rolling Shutter of the Canon C500 MarkII

Using our strobe light at 300Hz, we get a sequence of black and white bars due to the nature of read out of CMOS sensors. Each pair represents 3.33ms.

Seems like we are jumping from one fantastic result to the next – the rolling shutter at 6K RAW (5952×3140, 17:9) comes in at a strong 15.8ms – see figure 3 below.

This is the best result we have measured on a full frame sensor so far! For full frame sensors we haven’t seen a value below 20ms yet, 22ms was quite good already (e.g. SIGMA fp), while for instance the Panasonic S1H falls around 25ms.

Fig. 3: Canon C500 Mark II rolling shutter for 17:9 5952×3140: very good 15.8ms are measured.

Latitude (underexposure) Test of the Canon C500 Mark II

Latitude is the capability of a camera to retain detail and colors while over- or underexposing the image. As ususal, we exposed Nino’s face at max. 60% luma value on the waveform monitor, the white sheet of paper below the color checker comes in at around 65%.

This gives a base exposure of 360° shutter at f4.0 at ISO800 (UHD 25fps) – our 0 stops under baseline (internal noise reduction “off” for all tests). From there we successively underexpose the studio scene by increasing the shutter speed to 180°, 90°, 45°, 22.5° and finally 11.25° (5 stops underexposure).

Then, in post production (using DaVinci Resolve 16.1 studio) we push the underexposed images back to the 0 exposure value.

This test reveals a lot of insight about how well a camera (and the used codec) retains colors and detail. Most cameras break apart at 3 to 4 stops underexposure. Exceptions so far: BMPCC6K using BRAW 3:1 and Panasonic S1 (using ProRes HQ via an external recorder) which retain a usable image at 4 stops under.

Fig. 4: base exposure of our studio scene for the Canon C500 Mark II, using CLOG2 with the internal XF-AVC 10bit codec at ISO800.

Also here, using the internal 6K RAW results in much noisier images than the internal 10bit XF-AVC provides, so in order to have a better “out of the box” baseline when comparing these images to previous cameras, I decided to show you the internal XF-AVC 10bit images. And you won’t be disappointed ;-).

What goes without saying is that in the hands of an expert the 6K RAW files provide even better flexibility in post, especially the heavily underexposed images.

Let’s jump ahead and have a look at the image which is 3 stops underexposed, and pushed back in post – remember, this is the point where most of the previously tested cameras start to fall apart:

Fig. 5: The Canon C500 Mark II at 3 stops underexposure, pushed back to 0 using CLOG2 with the XF-AVC 10bit codec at ISO800. Still almost perfect.

And the Canon C500 Mark II does not disappoint – the results is almost perfect: very little noise is present, more as a fine grain which looks very good.

Fig. 6: The Canon C500 Mark II at 4 stops underexposure, pushed back to 0 using CLOG2 with the XF-AVC 10bit codec at ISO800. Noise kicks in but it stays finely dispersed. Very usable still.

More or less the same holds true for the image in figure 6: at 4 stops under, pushed back to zero the image is still quite usable. In addition, the internal XF-AVC encoder manages to encode the noise as finely dispersed grain, hence you can add noise reduction in post production and you get a very usable image – for examplethe noise in the lower right hand side of figure 6 cleans up well – see figure 7:

Fig. 7: The Canon C500 Mark II at 4 stops underexposure, pushed back to 0 using CLOG2 with the XF-AVC 10bit codec at ISO800. Noise kicks in but it stays finely dispersed – using noise reduction in DaVinci Resolve (3 frames temporal, 5 threshold value and 4 spatial threshold) it cleans up very well.

Finally, we are reaching now the 5 stops of underexposure. No camera so far has survived this torture test.

Fig. 8: The Canon C500 Mark II at 5 stops underexposure, pushed back to 0 using CLOG2 with the XF-AVC 10bit codec at ISO800. Very noisy, and horizontal as well as vertical stripes start to appear. Surprisingly good however.

Surprisingly, using noise reduction in post production leads to an image you may still get away with – noise is rather finely dispersed and not too distracting (however, faint horizontal and vertical stripes are appearing) – see figure 9 below:

Fig. 9: The Canon C500 Mark II at 5 stops underexposure, pushed back to 0 using CLOG2 with the XF-AVC 10bit codec at ISO800. Using noise reduction in DaVinci Resolve (3 frames temporal, 15 threshold value and 5 spatial threshold) it cleans up OK – still noticeable are vertical and horizontal stripes, but you could get away with this image.

Summary

The C500 Mark II clearly sets the benchmark for 2020 so far in our testing using a 6K Full frame sensor. Even when the second best option is selected, the internal XF-AVC 10bit codec, the results are still fantastic (internal 6K 12 bit raw requires more post processing).

A strong 13.1 stops of dynamic range, the lowest ever measured rolling shutter of 15.8ms for a full frame sensor, and clearly showing what is possible at the underexposure test: using the internal 10bit UHD XF-AVC codec basically down to 5 stops underexposure the image is still usable when applying noise reduction in post.

What are your experiences with the Canon C500 Mark II so far? Let us know in the comments below.

29
Leave a reply

guest
1
1
- 2
Filter:
all
Sort by:
latest
Rick Leads
Rick Leads
Member
10 days ago

Really enjoying the Lab test section. Would love to see the latest and greatest Arri results as that should be the gold standard. Also would like to see the new BM12K when you can get one.

I use the Original Canon C500 with the CD raw recorder/monitor and have been happy with results. Really like Canon color science. Also own the BM4KPC (global shutter is nice) and BMCC 1080P but it is very difficult to blend the early BM cameras with the Canon. Considering the BM12K (looks like better color now) or the Canon C500 MII once it is discounted. Higher res would be useful for stability options in post and always wanting for better dynamic resolution. That’s where lab testing is really helping. Any tools to cut through the hype will always be welcomed. Thanks!!

 Jospeh Slomka
Jospeh Slomka
Member
August 25th, 2020

Looking at the waveform from the XYLA chart it appears that step 11 is smeared all the way from step 10 to 12. While technically there from a dynamic range standpoint, it would not be a useful for photography. Step 12 looks even less useful and the half stop of 13 is not use able.

Is there an explanation on the difference between imatest and xyla ? I would rate this camera close to 10 and change stops based on the XYLA

 Luka Stojcic
Luka Stojcic
Member
April 19th, 2020

Did you guys have a chance to test the rolling shutter in the 4K Super-35 mode? Theoretically the result would be much better than the 5.9k full frame readout, correct?

 Luka Stojcic
Luka Stojcic
Member
April 23rd, 2020
Reply to  Gunther Machu

Gotcha. Thanks

simeone ricci
Guest
April 17th, 2020

the problem is only one.The cost

Pablo
Pablo
Guest
April 17th, 2020

Is there a test like this with the Varicam LT??

Member
April 16th, 2020

Thank you Dr. for the results! Test Lab is my favorite series of Cinema5D : ) I’ll be vary curious to see how the C500 mk II 6K sensor stacks up with the 1D X mk III 5.5K sensor.

After a week with the 1D X I noticed just how noisy the RAW footage is (and how clean compressed is!) and could never get clips as noise free in Resolve or Final Cut Pro without losing detail as the built in H.265/H.264 files. Which led me to think – does all in camera noise reduction inherently look better than using a NLE program since the NR algorithm is searching for a specific fixed noise pattern from a sensor’s circuitry versus a universal one? And if so would it be possible to have in practice a Noise Reduction ‘LUT’ so to speak from manufacturers?

Hoping Monstro VV, FX9, Venice, C700 FF & Alexa LF will make their way to the Lab at one point :D

Chris Clifton
Guest
April 16th, 2020

Thank you for this great review. Mine was delivered right before the Covid lockdown, so it’s great to know I have a top notch, fantastic camera to offer my clients once we all get back to work. Can’t wait! Thanks

Anthony
Guest
April 16th, 2020

Can you please test the Panasonic EVA1, too? Plus you should make the list of links to all your tests and results easier to reach because they are extremely thorough and useful and now they are spread over the years

Jamie Walter
Jamie Walter
Guest
April 15th, 2020

Is there a database for all your results? You mention the chart but I don’t see a link?

 Matthew Thomas
Matthew Thomas
Member
April 15th, 2020

It’s frustrating to have what would otherwise be the right camera for my needs hobbled by the use of EF or PL mounts rather than the much more versatile and future-proof RF mount. I know many existing workflows are not using RF, but Canon’s own brilliant RF to EF adaptors solve 99% of these issues. I’m outright confused at how and why Canon keeps hobbling their products in such unnecessary ways.

Dalton McRowan
Dalton McRowan
Guest
April 16th, 2020
Reply to  Matthew Thomas

Canon also doesn’t let you control the focusing motors in their own lenses. It’s ridiculous to have to bolt a mechanical contraption onto a lens for follow-focus, when the lenses have focusing motors built in.

Why? Because they want to sell you $5000 cinema lenses.

Tom Roper
Tom Roper
Guest
April 17th, 2020
Reply to  Matthew Thomas

Just so I understand you, EF lenses are brilliant when you need a Canon adapter but hobbling when you don’t. The horror.

BOUNCE
Member
April 15th, 2020

No Reds tested yet? I would be curious to see some independent test.

 Abe Halpert
Abe Halpert
Member
April 16th, 2020
Reply to  BOUNCE

Especially Komodo when it comes

Florian Lindenberg
Florian Lindenberg
Guest
April 15th, 2020

+1 for Ryan and FX9. And also Mini LF and especially Monstro would be really interessant see in comparison! :)

Nick Lam
Guest
April 15th, 2020

Could you guys do a cumulative chart of all of the cameras you’ve tested thus far for DR and Rolling Shutter (ms)? It would be nice to have a dedicated section where we can just go to and reference at any time.

Misha Engel
Misha Engel
Guest
April 15th, 2020
Reply to  Nick Lam
Johnnie Behiri
Admin
April 16th, 2020
Reply to  Nick Lam

Hi Nick.
We are working on this. It will take a bit of time, but it will come…

Thank you
Johnnie

Ryan Rourk
Guest
April 15th, 2020

Would love to see the Sony FX9 tested as well.

Falko
Falko
Guest
July 8th, 2020
Reply to  Gunther Machu

when :) I would be delighted

1
1
- 2
Filter:
all
Sort by:
latest

Take part in the CineD community experience