Description image

Canon C300 Does Great in Low Light but White Wall 'Breaks' the 8-Bit Codec

01.2.12 @ 10:10AM Tags : , , ,

Here’s an excellent test/review of the Canon C300 camcorder by Paul Steinberg. He shoots a number of low light shots with the camera but also manages to “break” the 8-bit codec, in his own words. It’s hard to make out what is C300 compression and what is internet h.264 compression, but in Paul’s words “you can see a ton of quantizing little blocks” — even when viewed on a TV. No matter how good your 8-bit implementation is… it’s still 8 bits. Is this a deal breaker for you?

Deal breaker or not, the C300 also looks to be the best low-light camcorder on the planet. 20,000 ISO is not something most people will be using regularly, but it’s there and useable (in some situations). 5,000 ISO looks good, and in practice more people will notice (as a positive) the ISO flexibility than will have a problem with the 8-bit codec. Paul also notes that the camera does not seem to take well to pushing the exposure in post in terms of shadow detail — don’t be afraid to use the higher ISO settings.

[via FreshDV]


We’re all here for the same reason: to better ourselves as writers, directors, cinematographers, producers, photographers... whatever our creative pursuit. Criticism is valuable as long as it is constructive, but personal attacks are grounds for deletion; you don't have to agree with us to learn something. We’re all here to help each other, so thank you for adding to the conversation!

Description image 31 COMMENTS

  • i read somewhere that 20000 iso @C300 is the same as 6400iso @ canon DSLR.. is that true?

    • Andreas Lange on 01.2.12 @ 11:52AM

      No its not.
      6400iso on a canon EOS 60D is way more noise than 20 000 on the C300.Especially when it comes to those colors.

      • he’s actually referring to light sensitivity, not noise: I’ve also read that you get an image just as bright with either 20000 iso @C300 or 6400iso @ canon DSLR (I think it was 1D4 that they were testing against)

        yes, 20000 ISO on the C300 may be a lot cleaner than high ISO on any DSLR, but if it’s not any brighter than 6400 ISO on a DSLR, we should know: I’d also like to hear more about this issue

        ISO is supposed to be a standard, but not everybody follows those as nicely as they should…

        • +1.

          I’ve wondered about this; now that the ISO system is entering the video market (as opposed to gain in dB) I’m never quite sure if it’s used as it should be, or if companies are just throwing the term around. Could really do with a solid test of such things. Perhaps Phillip Bloom would be the guy for the job…

          • If thats the case it’s going to make setting exposure with a light meter pretty hard.

          • I personally don´t like that bullshit that those modern companies make.. why don´t they measure the amount of light that the senzor gains and assign a standard iso value according to some world standard iso chart or whatever ..

            ISO 160 on GH2 gives you the same amount of light as 5DmII@ iso 400 (it was proven by Andrew Reid on and the C300 at it´s “gorgeous” 20 000iso is only a 6400iso on a 5D so what is the real iso performance? is it another marketing trick? is then the usable 6400iso on C300 equvalent (i mean light gain.. not iso numbers! ) to iso 3200 on GH2? i´m not a fanboy but you can have a clean image on GH2 @3200iso when hacked with high bitrate so what is so gorgeous about that C300? at ISO performance it will be outperformed by 5DmIII , nikon D4, 1Dx, D800 and maybe even the GH3 when we would compare iso sensitivity not according to number but light that we gain..

  • Wow. That last shot with the beach…

    I really like this camera. I wonder what that white wall would look like after it’s run through Magic Bullet’s Denoiser plug-in. I once shot a test of my wife in the living room at 3200 ISO and it did a great job at cleaning all that blockiness.

  • Well mathematically the 8 bits has to break when dealing with light and shading. The math is will always be wrong because no matter what woo woo tricks you do.

    And Canon released a whitepaper explaining the strategy behind its sensor which answers a lot of questions:

    Its using lots of clever “tricks” such as: When shooting 24p you are actually getting a 1/60th shutter to decrease rolling shutter, The noise reduction secret is Canon increases the noise early so the noise reduction inside the camera has a bigger chunk to work with so when you use higher ISO its cleaner.

    Thats why in the video is noisier at 800 ISO than at 5,000

    Again, clever…but doesn’t change the post production mathematics.


    “The design strategy of this new sensor is to not to seek any form of “4K” resolution — but rather to specifically confine the reconstruction of each of the R,G, and B video components to a full digital sampling structure of 1920 (H) x 1080 (V) — according to the SMPTE 274M HDTV Production Standard.”

  • he graded it at8bits or transformed the vid to 10bits first? most of the time a 8bits to 10bits conversion can solve this kind of coded breaking.

  • Wilbert Thomas on 01.2.12 @ 6:58PM

    The biggest problem with the c300 is the price, not the 8 bit codec

  • Wilbert Thomas on 01.2.12 @ 10:40PM

    Interesting, DP Rodney Charter claiming the c300 is the “Camera of the Decade”. Everybody has an opinion, we should just use the camera that best for the job we need it.

    • That is interesting indeed. I agree with him that it could be the camera of the decade — we disagree about WHICH decade, however. The C300 would definitely be the camera of 2000-2010, were it out… but I doubt anyone will be saying the same thing in 2020.

  • Not to diminish the test, but Paul had the camera for a day – as he mentioned on CML (cinematography mailing list). Must admit, I’m not a big fan of the sensationalist heading of the article. A lot who have actually tested the camera have grown to love the images coming from it. And, as with any equipment – be it camera or others, it’s a tool for a specific execution. Finally, as with any tool, it performs differently in different hands.

    Happy New Year Koo.

  • Paul Steinberg on 01.3.12 @ 11:08AM

    I’ll agree with Mai, the description of this blog post leans a little sensational. If you listen to my description you’ll hear how the camera arrived suddenly and was whisked away just as quickly. This was no where near the extensive testing that others are currently putting it though when they have it for longer periods.

    As a Red owner with a bunch of PL mount glass my ‘production’ work tends to go to that camera, I am looking at the C300 to replace my 1DmkIV for work that I currently shoot on a DLSR and when shooting DSLR I keep the camera stripped down, shoot with EF mount zooms and use the camera’s small size to keep moving quickly and that is how I worked with the camera that day.

    As far as breaking the codec on the one shot, I do explain that the compression artifacting was seen clearly when standing in a professional post-colour suite and staring at the image on very large, very accurate monitors. Depending on your final output, you may never see it. If I had to shoot that shot again now knowing what I have learned from the ISO tests I did at the end of the day, I would have increased the ISO to get more information in the middle of the LOG.

    I hope anyone reading this will also note that that was the only shot of the day that had any issues whatsoever. Moire is essentially eliminated, Jellocam is essentially eliminated. Overcrank works the way it should (albiet in 720 mode). This was a pre-release version of the camera and it worked flawlessly on a December day in Toronto all with a single little battery. The files worked flawlessly on multiple systems and didn’t require any software updates or special hardware.

    I don’t know if I’d call it the ‘camera of the decade’ but it is a massive step forward from DSLR shooting and with the new EF mount Cinema Primes/Zooms that canon are bringing out it is clear that they are committed to giving cinematographers the tools to work they want for this camera platform.

    As a Red One, soon to be Epic owner I intend to buy a C300 instead of a Scarlett once they are available.

  • lets get it straight this is NOT “GRADING” its “COLOR CORRECTION’ and there is a big difference. if you don’t know the difference, look it up please. I hate how everyone slaps the grading word on color correction to sound like they are doing something special. they aren’t.

    that said this test was a pretty big yawner on several levels. please, lets see some base comparisons here with a couple other cameras for reference point. the candle light tests where just killed by vimeo’s re-compression for the most part.

    with the beach shot, w/o seeing an iso 320/640 shot, how dark was it really ? can’t tell, useless.

  • I haven’t seen a single test that hasn’t drawn virulent outbursts from the increasing lot of cinematograpy/videography Talibans who seem to believe that no one should state an opinion unless it is agreeable with their terms. That said, let me declare that I am an amateur with a DSLR learning a lot from various opinions that professionals who owe me no duty to explain anything to me, share with us. I appreciate the differing opinions, it alerts me on what to look for when I decide to go for an upgrade later. Thanks Koo & Paul and the many other pros that educate us daily free of cost.

  • I disagree with the notion that the 8-bit limitation causes quantization blocking. It is MPEG-2′s ancient inefficiency. 8-bit certainly limits precision when compressing, and 10-bit codecs for example, eliminate some of this, by providing a higher range of calculation when doing quantization, but by no means is highlight blocking excusable on a high bitrate codec.

    Take any well-encoded H.264 Blu-ray disc – no, that’s unfair because it’s controlled two-pass encoding – let’s take quality based (CRF) x264 encoding in 8-bit. At sufficient bitrates, that codec would laugh at such a shot. This particular H.264 implementation is above the DSLRs, but consider the Canon 5D’s 48Mbit implementation, versus Panasonic’s 21Mbit. Actually, while we’re at it, compare that same nearly-50Mbit bitrate to the C300′s 50Mbit MPEG-2.

    I can understand cost here – H.264 encoding hardware is likely expensive, or at least high quality implementations are. Software implementations, such as x264, are not designed to run on standalone hardware, but I sense that if they wanted the cheaper route, they should have compressed material to a computational and economically cheap format. They’re busily paying MPEG-LA for H.264 licenses, so they make the best of it…

  • Can anyone explain to me why I should buy the C300 instead of the Sony F3 and a KIPro Mini?

    An F3 will run you $13,160 (currently w/ rebate) and, if you add the AJA KiPro Mini for $2000 and some proper CF cards, that comes to around $15,500 depending on the card capacity you choose to buy.
    With the KiPro Mini, you can record a glorious 10bit 220Mbps Apple Pro Res HQ file that drops right into FCP. No need for Magic Bullet’s Denoiser or any other little bandaids to try and remedy a crippled 8 bit codec.

    I just don’t get the Canon C300 hoopla and all the defense of it’s 8 bit system. If you’re going to shell out $14K, wouldn’t an extra $1500 for a 10bit 4:2:2 220Mbps codec be warranted? And the F3 has real XLR connections, not to mention the ability to record full 1080P 60FPS or true 4:4:4 to an appropriate recorder when needed.

    As I’ve stated in a different thread, I am a Canon fan as I have several Canon DSLR’s and a compliment of Canon glass but, I just don’t understand this camera being a prudent choice unless it’s diminutive size is your main criteria.

  • Dixter – now the Nikon D4 is out at $6k in feb – with totally sweet clean HDMI out – just add a ninja for $1500

    So D4 + ninja for $7500 … bonus the D4 also has 3 shooting formats on its chip – FX + DX + 1:1 crop all in Full HD.

    I’ll wait for tests first however as the C300′s chip has been reported to be very nice for skin tones with the way the image hits the chip and is reduced down by 4 etc etc.. i like this idea for Moire and dof. and diag lines etc etc.. Moire is very important to me as I do interview/doco style stuff..

  • Paul Steinberg on 01.7.12 @ 9:21AM

    Dixter, 8-bit 50Mb 4:2:2 is identical to every XDCamHD camera out there and the Sony F900R’s are 8-bit colour as well. They are very well established cameras and my point is just that 8-bit colour is acceptable for many applications. For my day with the C300 there was only one shot where the limitations of the codec were visible and I was trying to set up shots that would uncover flaws so all in all I would say it held up fine. The only difference between 8-bit and 10-bit would be in post colour grading where you could push the 10-bit harder before artifacting was introduced.

    For me the big difference between F3 and C300 is PL vs EF glass and being self contained vs shooting with a Mechano set.

    I should mention that I am camera agnostic. I’m not partial to any manufacturer, I just look for the best took for the job. I’m looking at C300 to replace the type of work that I am currently shooting with my 1DmkIV so I need a self contained camera that utilizes EF lenses. The C300 has XLR and BNC in/out like the F3 but if you’re going 4:4:4 S.log to an SR recorder with an F3 you might as well just get an Red or Alexa. I’ve stopped getting hung up about data and sampling rates when I first tested DSLR cameras. Now I test the cameras myself because these engineers are making some brilliant images when the numbers on the sheet say that it shouldn’t be looking so good.

    Tony, I can tell you that with my limited time with the camera I found that moire was essentially eliminated, jellocam was essentially eliminated and fleshtones held up in shots where they would typically go plastic with a DSLR. Very encouraging.