Canon Cinema EOS C300 Mark II Lab Test

[Canon Rumors Guy]

<p>The folks at Cinema5D have completed their first lab test of the Canon EOS C300 Mark II. One of the big upgrades was the 15 stops of dynamic range that Canon has claimed the new EOS C300 Mark II has. Cinema5D tested this in their lab and found that the new camera has about 2 stops less of dynamic range than they were expecting. The dynamic range appears to equal that of Sony’s FS7, but is well short of the ARRI ALEXA. Cinema5D did prefer the overall image quality of the EOS 300 Mark II after comparing it to the Sony FS7.</p>
<p><strong>From Cinema5D:</strong></p>
<blockquote><p>Even though the Canon C300 Mark II seems to be quite on par with the Sony Fs7 in terms of dynamic range, I must also say the image of the C300 Mark II is more neutral and the noise is less saturated and mushy in comparison to the Sony FS7. So while unfortunately there is a lot of noise in the shadows, at least it doesn’t look so bad. Also the C300 Mark II seems to be about 1 stop more light sensitive than the FS7 at high ISO speeds. <a href="https://www.cinema5d.com/canon-c300-mark-ii-review-dynamic-range/" target="_blank">Read the full lab test</a></p></blockquote>
<p><strong>Dynamic Range Comparison:</strong></p>
<div id="attachment_22633" style="width: 738px" class="wp-caption aligncenter"><a href="http://www.canonrumors.com/wp-content/uploads/2015/09/dynamic-range-C300-mark-ii-vs-fs7-vs-alexa_B.jpg"><img class="size-medium wp-image-22633" src="http://www.canonrumors.com/wp-content/uploads/2015/09/dynamic-range-C300-mark-ii-vs-fs7-vs-alexa_B-728x271.jpg" alt="Click for larger" width="728" height="271" /></a><p class="wp-caption-text">Click for larger</p></div>
<p><strong>Preorder: </strong><strong><a href="http://www.bhphotovideo.com/c/product/1134579-REG/canon_0635c002_eos_c300_mark_ii.html/BI/2466/KBID/3296" target="_blank">Canon EOS C300 Mark II EF</a>| <a href="http://www.bhphotovideo.com/c/product/1134580-REG/canon_0635c009_c300_markii_cinema_eos.html/BI/2466/KBID/3296" target="_blank">Canon EOS C300 Mark II PL</a> Rent: <a href="http://www.lensrentals.com/rent/video/cameras/canon/canon-eos-c300-mark-ii-dual-pixel-ef" target="_blank">LensRentals.com</a></strong></p>

 

[gsealy]

It's a really good test of the C300-II. Consumers need this info as things are changing so fast and the prices are all over the place.

 

[privatebydesign]

I have no interest in video or the C line, but taking a couple of seconds to read his methodology pointed out some obvious issues.

He tested the C300 MkII in 4K, in 4K it records in 10 bit, in 2K it records in 12 bit. Now he goes on to say he couldn't see a difference and felt 4K downsampled looked better, but that isn't the point of a lab test, the point is to remove the subjectivity and just record results.

 

[jeffa4444]

Regardless of whether you record 2K at 12bit or 4K at 10bit the fact that the camera only records around 12 stops of DR when they state 15 stops is the issue right there.
Arri Alexa uses large pixels therefor is able to gather more light and Arri have constantly improved the firmware to further refine the performance. In testing Sony cameras we find they have better resolution than the Alexa / Amira but that is not nessisarily what the cinematographer is looking for although VFX would be happier. Canon seem from this test to have used Sony as their benchmark not Arri maybe that is coming with their 8K camera (although hopefully its a. not 16x9 but full frame and b. not in a C300 format body but more like the Alexa). Arri have been making TV & cinema cameras since the thirties unfortunately the Japanese DONT listen to the end user in cinematography in the way Arri does and Arri will have its own Alexa replacement in 2016.

 

[Etienne]

Regardless of whether you record 2K at 12bit or 4K at 10bit the fact that the camera only records around 12 stops of DR when they state 15 stops is the issue right there.
Arri Alexa uses large pixels therefor is able to gather more light and Arri have constantly improved the firmware to further refine the performance. In testing Sony cameras we find they have better resolution than the Alexa / Amira but that is not nessisarily what the cinematographer is looking for although VFX would be happier. Canon seem from this test to have used Sony as their benchmark not Arri maybe that is coming with their 8K camera (although hopefully its a. not 16x9 but full frame and b. not in a C300 format body but more like the Alexa). Arri have been making TV & cinema cameras since the thirties unfortunately the Japanese DONT listen to the end user in cinematography in the way Arri does and Arri will have its own Alexa replacement in 2016.

The Canon C300 and the Sony FS7 are affordable to a lot of independent documentary and film makers. The Arri, not so much.

 

[privatebydesign]

Regardless of whether you record 2K at 12bit or 4K at 10bit the fact that the camera only records around 12 stops of DR when they state 15 stops is the issue right there.
Arri Alexa uses large pixels therefor is able to gather more light and Arri have constantly improved the firmware to further refine the performance. In testing Sony cameras we find they have better resolution than the Alexa / Amira but that is not nessisarily what the cinematographer is looking for although VFX would be happier. Canon seem from this test to have used Sony as their benchmark not Arri maybe that is coming with their 8K camera (although hopefully its a. not 16x9 but full frame and b. not in a C300 format body but more like the Alexa). Arri have been making TV & cinema cameras since the thirties unfortunately the Japanese DONT listen to the end user in cinematography in the way Arri does and Arri will have its own Alexa replacement in 2016.

I am not making claims for anything other than the fact that the tester, while presenting his test as a scientific style lab test, took the science and legitimacy out of the test when he used subjective decisions in not shooting like for like.

Where does Canon say the C300 MkII shoots 15 stops of DR in 4K at 10 bit? They don't, even the tester says "I am aware Canon says they achieve best image quality at 2K in 12 bit. " So in an outright test of DR the tester arbitrarily decided to not use the settings the camera maker says would give the best output, and then says the output isn't what he was expecting! Further, where do Canon lay out their definition of DR? Where does the tester lay out his definition of DR?

I agree it looks like the C300 MkII is incredibly close to the Sony and the tester says he preferred the output of the Canon over the Sony, but again that is subjective, and both were shot in 10 bit and being compared to a camera shot in 12 bit! Hmm, wonder if that lack of bit depth has an impact in IQ? Canon say it does............

 

[Seb-cinema5D]

Hi there,

I tested (measured) the DR in 2K 12 bit. The results were the same. In the article I decided to print the 4K version because I felt the grain looked nicer and the emphasis was on comparing C300 vs FS7.
Cheers

 

[Diltiazem]

Hi there,

I tested (measured) the DR in 2K 12 bit. The results were the same. In the article I decided to print the 4K version because I felt the grain looked nicer and the emphasis was on comparing C300 vs FS7.
Cheers

Why did you use different apertures for Canon and Arri?

 

[JimMartin]

Okay, I think this has gotten a bit out of hand....I'm gonna say I'm 99.9% certain that Cinema5D had one of the prototype cameras....ones that were built in March, not a production run camera. The first production run cameras were just delivered at the end of last week here in the US and I don't think any have been delivered in Europe (where they are located) yet. What they are talking about in terms of the black noise was on 3 different prototypes that I had my hands on. They had good range but we didn't bother testing it because we knew it would not be accurate compared to the final, production run version. I know there is a TON of tweeks, adjustments, etc that have been made to the camera to get it ready for release. The bottom line is, I'd put a lot more credibility in the ASC DP, Gale Tattersall (who shot the demo film "Trick Shot") who says there are definitely 15 stops vs. Cinema5D's test.

 

[privatebydesign]

...
Where does Canon say the C300 MkII shoots 15 stops of DR in 4K at 10 bit? They don't, even the tester says "I am aware Canon says they achieve best image quality at 2K in 12 bit. " So in an outright test of DR the tester arbitrarily decided to not use the settings the camera maker says would give the best output, and then says the output isn't what he was expecting! Further, where do Canon lay out their definition of DR? Where does the tester lay out his definition of DR?
...

http://cinemaeos.usa.canon.com/wp-content/uploads/2015/04/EOS_C300MKII_Specs_4815.pdf

You're right, Canon doesn't qualify its "15 stops of dynamic range" with which resolution it applies to, therefore it is reasonable to assume that it applies to both.

Not at all, I read the words "up to 15 stops of dynamic range," . Surely that means not all the time......

 

[privatebydesign]

Okay, I think this has gotten a bit out of hand....I'm gonna say I'm 99.9% certain that Cinema5D had one of the prototype cameras....ones that were built in March, not a production run camera. The first production run cameras were just delivered at the end of last week here in the US and I don't think any have been delivered in Europe (where they are located) yet. What they are talking about in terms of the black noise was on 3 different prototypes that I had my hands on. They had good range but we didn't bother testing it because we knew it would not be accurate compared to the final, production run version. I know there is a TON of tweeks, adjustments, etc that have been made to the camera to get it ready for release. The bottom line is, I'd put a lot more credibility in the ASC DP, Gale Tattersall (who shot the demo film "Trick Shot") who says there are definitely 15 stops vs. Cinema5D's test.

Thanks Jim, very interesting insight. I am watching Trick Shot as I write this..........

 

[Local Hero]

Okay, I think this has gotten a bit out of hand....I'm gonna say I'm 99.9% certain that Cinema5D had one of the prototype cameras....ones that were built in March, not a production run camera.

No way.
Down in Australia they have been shipping to clients for a couple of weeks.
The promo footage, shot in the Greek Islands, months ago was on a full production release camera.

 

[bgoyette]

Okay, I think this has gotten a bit out of hand....I'm gonna say I'm 99.9% certain that Cinema5D had one of the prototype cameras....ones that were built in March, not a production run camera.

No way.
Down in Australia they have been shipping to clients for a couple of weeks.
The promo footage, shot in the Greek Islands, months ago was on a full production release camera.

Actually, no Brett Danton shot that on a prototype with the early firmware. Production cameras were first seen at IBC and began shipping in tiny numbers last week. That said, from the photo that Sebastian posted in his article, it would appear to be the production model, unless he retouched the hell out of it.

 

[bgoyette]

And anyone that was expecting the "15 stops of DR" to mean new sensor tech to be trickled down to DSLRs can forget it. From the review:

"An additional surprise was that I saw a lot of noise in the dark areas on the Canon C300 Mark II even at its base ISO of 800. While the camera has a very clean image in the brighter areas and has a really neutral tone with minimal color noise, there is a point in the dark areas where the noise kicks in strongly. When I did some test shots of natural subjects I realized that images shot on ISO 800 that are just 1 stop underexposed can quickly become a little too noisy for my taste especially in comparison to a camera like the Arri AMIRA (our reference camera). That is rather the opposite of what I expected after all the talk about the Canon C300 Mark II sensor and processing being so new and advanced. Ultimately one has to admit the 5 year old Arri ALEXA sensor is doing better…"

The part that bugs me about Sebastian's review is that he keeps saying this is a comparison of the Fs7 with the C300 mark II, and then he makes a statement like this. First he switches to comparing it to the Amira (he does that whenever he doesn't want to talk about the FS7 I've noticed), and here's why -- In Slog3, the FS7 shows pretty much the same noise in the shadows...Slog3 and Clog2 are similar in that they are lifting the shadows quite a bit, and thus they expose the noise in the deep shadows that you'd never see once you graded it. Someone in his position knows that you overexpose logs built like this, and you underexpose only if you have no idea what you're doing. He's acting surprised here, but he knows better. He also really fails to smell the coffee in his 12,800 low light test. While he mildly states that it's a little better than the FS7 in this regard..(it's a mile better)...(let me pull a sebastian -- I'm doubtful the Alexa/Amira would even register much of a signal at 12,800.) Put that chart up against any other camera (short of maybe the A7S) at that ISO and you'd think a he would be impressed with what the C300 is delivering on the lowlight front.

 

[Johnnie]

Guys, just to clarify, The lab test was done with a production camera. We never do those tests on a pre production models or cameras not having final firmware.

Thanks!

Johnnie- cinema5D

 

[LetTheRightLensIn]

Okay, I think this has gotten a bit out of hand....I'm gonna say I'm 99.9% certain that Cinema5D had one of the prototype cameras....ones that were built in March, not a production run camera. The first production run cameras were just delivered at the end of last week here in the US and I don't think any have been delivered in Europe (where they are located) yet. What they are talking about in terms of the black noise was on 3 different prototypes that I had my hands on. They had good range but we didn't bother testing it because we knew it would not be accurate compared to the final, production run version. I know there is a TON of tweeks, adjustments, etc that have been made to the camera to get it ready for release. The bottom line is, I'd put a lot more credibility in the ASC DP, Gale Tattersall (who shot the demo film "Trick Shot") who says there are definitely 15 stops vs. Cinema5D's test.

When has the DR ever differed much between pre-pro and production bodies? So far the banding and DR from the preo-pro DSLR from Canon have always been at least as good as from the production models, going back years.

 

[FunkyCamera]

Who cares if it doesn't actually have 15 stops of dynamic range in normal situations? It's been demonstrated here time and time again that dynamic range is just a pointless marketing gimmick for people who enjoy pleasuring themselves to 200% zoom crops instead of real photos or videos. Leave that to the people buying Sony junk and just be happy that Canon still make the best kit out there for real people.

 

[Tugela]

Who cares. Not like real people could afford this camera anyway.

 

[sanj]

Who cares if it doesn't actually have 15 stops of dynamic range in normal situations? It's been demonstrated here time and time again that dynamic range is just a pointless marketing gimmick for people who enjoy pleasuring themselves to 200% zoom crops instead of real photos or videos. Leave that to the people buying Sony junk and just be happy that Canon still make the best kit out there for real people.

I care. I shoot in normal situations day in and day out.
Who has demonstrated that dynamic range is a pointless gimmick??????
Dynamic range has NOTHING to do with crops. lol.

 

[Local Hero]

Actually, no Brett Danton shot that on a prototype with the early firmware. Production cameras were first seen at IBC and began shipping in tiny numbers last week. That said, from the photo that Sebastian posted in his article, it would appear to be the production model, unless he retouched the hell out of it.

Nope.
The camera Brett Danton used was at the SMPTE show in Sydney.
Was on the show floor on the 13th July.
Identical to full production camera in every way.
Canon may have called it a Prototype camera so that sales clients didn't get their nose out of joint that some people and not them were being allowed to shoot on the camera.

Actual clients in Australia had cameras in their hands two to three weeks ago (EF only, no PL cameras yet).
Batteries are in really short supply in Australia though.
People always get their nose out of joint that others have received their cameras before them, so it's safer to just say they have not shipped yet to clients who haven't received theirs yet.

Nothing against the C300 MK II.
I think it is great.
These tests are pretty indicative of the dynamic range of this camera.
But there is no magic 15 usable stops in ANY camera.
Even the F65 doesn't have 15 usable stops.