Description image

8-Bit is Still 8-Bit, Why DSLRs Are No Match for the 12-Bit Blackmagic Cinema Camera

People have done as much as possible to argue against the test that was performed with the Blackmagic Cinema Camera by Marco Solorio and OneRiver Media. That test compared the 5D Mark III and the BMCC, and even though the superior resolution and dynamic range of one of those cameras should have been obvious, many still prefer the Mark III and complained that a better picture profile or some post sharpness would make the differences less noticeable. Well, Marco is back with a new test, and (in my personal opinion) it’s hard to argue against these results, as he’s pulled out all of the stops to really test the Blackmagic Cinema Camera and show why 8-bit will always be 8-bit, and what it means when you’ve got the capability for 12-bit images.

Here is the description below the video:


Going a step closer with the Cinema Camera, Marco Solorio of OneRiver Media focuses on how 12-bit RAW compares in relation to 8-bit alternatives, including many camera solutions costing much more. The difference between “perceived dynamic range” and “available dynamic range” is explored in detail. Also covered is day-for-night, detail versus sharpness, and much more.

You could download the original 1080p file as Marco mentions, but I actually watched the compressed online version first, and the differences were still obvious.

As far as dynamic range is concerned, there is only so much you can do with 8-bit to protect the highlights. This was made clear to me in the Zacuto Revenge of the Great Camera Shootout 2012 test. I was able to see that test on the big screen, and when the DSLRs came up, it was clear that there was only so far they could be pushed before they started falling apart. The parts of the image that were massively underexposed to protect the highlights began to look grey and lifeless when they were brought back up. That’s what happens with 8-bit. There is only so much information contained within the image, and you can’t recreate something that doesn’t exist in the first place.

The bit-depth simply refers to the different combinations of ones and zeros that is possible (with color images having three channels of bit-depth). The more bits and values possible, the better the final quality will be.

  • 8-bit   – 256 different values
  • 10-bit — 1024 different values
  • 12-bit — 4096 different values
  • 14-bit – 16,384 different values
  • 16-bit – 65,536 different values

Some of you might already be saying that the CinemaDNG RAW files from the Blackmagic Cinema Camera take up a ton of space. That is absolutely true, and short of converting to a compressed RAW format like Cineform, it’s something to consider. If you notice, though, there is still a large difference between 8-bit and 10-bit in terms of the values possible, and if you shoot ProRes or DNxHD, you’ll be getting a 10-bit file. That is why even in compressed mode, you will still be able to technically pull more detail and color information out of the BMCC as opposed to all DSLRs which only output 8-bit 4:2:2 at their very best.

Internal processing bit-depth is another factor to consider. It’s why even though DSLRs are crippled in many ways, especially by the 8-bit color space, that they can still make beautiful images. The detail you are working with is usually starting from a 14-bit source, which is what most of the high-end DSLRs are capable of. After going through this processing, the final image has to be compressed once more to the final output, which for video, is 8-bit. So technically if you could pull RAW video data from these DSLRs, you could be working with a far better file, even if they are pixel-binning or line-skipping. It’s one of the reasons people seem to like EPIC/SCARLET footage over RED One MX footage at the same resolution even though they have the same sensor, the RED One is a 12-bit camera internally, while the SCARLET and EPIC cameras are both 16-bit internally.

Now, what about resolution? Adding sharpness does exactly what it’s supposed to do: make the edges more defined. I’ve never been a fan of sharpening, it might make an image pop a little more, but it’s not the same as actually having a lot of resolution. You can’t make a 720p image look like a 1080p image by sharpening, even on the web. That’s why I wanted to watch just the online version first, because it’s clear to me that detail and dynamic range do make their way through to compressed video, regardless of what many people might tell you. You will have to make sure that you are getting the least compressed file online as possible, but I consistently see better looking online footage from RED, the Arri Alexa, and the Sony F65 that is clearly superior to DSLR footage.

Of course, cost is a major factor. You might not be able to afford one of these high-end digital cinema cameras (even as a rental), and that’s what makes the Blackmagic Cinema Camera so special. For all its quirks, nothing can touch this quality at $3,000. Is this saying that everyone should go out and buy the BMCC? Absolutely not, and for many people, it’s going to be the wrong camera for what they’re doing, even if it does have higher resolution and dynamic range. For one thing, it can’t see in the dark like many DSLRs. It’s also going to need a battery of some kind, which is a far simpler addition on a DSLR.

Your DSLR can still give you a wonderful image. There is no doubting that, and if you’ve got one, don’t necessarily get rid of it just because a new camera is coming along that should technically outperform it. Many have stated that they like the footage coming from DSLRs better (I’m not one of those people). I thought this was an important topic to correct some of the misinformation floating around. You should not be ashamed of using a DSLR, even professionally, because in the end, if the client is happy and you’re making money, who cares? But there is very real evidence that you can do a lot more with the footage coming from the BMCC, and it’s going to show through even in a compressed online video.

Now, as for what your monitor can actually display, that’s another story completely, but we’ll have to save that for a future post. Let’s start a real discussion about this, because it’s important not just from a technical standpoint, but from an artistic one. If you’re limited technically by your camera, there is only so much you can do to compensate, no matter what anyone tells you, and in certain instances, it can make a huge difference.

What do you guys think? For those who wanted to see the Mark III with some sharpening, what do you think about that footage now compared to the BMCC?

Links:

Related Posts

  1. Blackmagic Cinema Camera Shows the Canon 5D Mark III Who's Boss
  2. We Got Your Dynamic Range Right Here: Blackmagic Cinema Camera Put to the Test
  3. Take an In-Depth Look at the Blackmagic Cinema Camera

COMMENT POLICY

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 149 COMMENTS

  • Great blog and an important one. I hope it helps some of those unaware understand the incredible value in the BMCC. There are as many cons to the sensor size debate as there pro’s when it comes to making low budget cinema imho.

    I think the BMCC is a truly amazing offering. I noted someone said that people won’t use this for cinema? I for one would be ver surprised if there isn’t a lot of micro budget indie cinema shot with this camera, as it quite simply exceeds production value for cost ratio more than has ever been possible before.

    Would have given an arm and a leg for it times gone past. Amazing days we live in… amazing days.

    • I said that, and what I wanted to say, is that Hollywood won’t use the camera for their films. I agree that this camera is a dream come true.
      What I was complaining about was the test itself. I felt the test was designed the wrong way.

      The camera is great. But there was a problem with the test.

      • Of course “Hollywood” will… anyone actually doing work in “Hollywood” would see that coming from a several miles away.

        Not sure how it’s not obvious even as an outsider.

      • There will always be flaws in any test, even the Great Camera Shootouts, with their extravagent budgets and countless man hours of work, people will always find flaws in the methodology of the tests. There will never be a perfect camera test.

        Camera tests, and in particular these One River tests, show the methodology the filmmakers use. The One River guys set up their tests to compare against their workflow with 5d Mk 3, which until the BMCC had been their workhorse. They find the BMCC to be superior for what they do. You can find many people on these boards who looked at the tests and found the 5d Mk 3 to be superior, it all matters how you use your camera and technology.

        • I remember reading about Miami Vice, one of the first big digi pictures, and Dion Beebe said he wasn’t sold on digital until he saw these big vistas of clouds, and how the clouds seemed to pop right out of the sky like 3d. To me that is testing. One River looks at the BMCC and it pops for them. We don’t have to agree with their views, but we should be thankful they are sharing their info, even if we view their methodology as flawed. That way when we go on to do our tests, we can build upon what they have shared.

    • I own a sony F55 – Stats: 16 bit raw, S-gamut, Slog, global shutter – and absolutely superior in every way

      EXCEPT ONE: Price

      Yes, the F55 is a much better camera than the BlackMagic Cinema Camera, in the same way that 12 is better than 8 bit, 16 is better than 12…. well, thats not even true… because it is exponential and not linear it means it’s a lot better for post processing.

      BUT – how much do you get for 3k dollars vs my 29,000 dollar camera. Is my F55 ten times better? To those who have the budget it sure is- like everything else in the world there is a point where you have to draw the line and say “THIS IS GOOD ENOUGH”

      I own a 5D2 and 5D3 – and I love those cameras a lot – and I used to argue until I was blue in the face.. and in the end of the day I now realize how much I was embarrassing myself. It’s like the difference between science and religion – while you can have a difference between religions – hard science/mathematics are not up for debate – the sun doesn’t revolve around the earth and 5 is half of 10…. and so is it with specs of a camera… those are the parts you just cannot argue with. That isn’t to say you can’t shoot a better movie on a 5D3 than a BMCC or F55, it just says that it’s an inferior camera. Don’t worry, this is not an attack on you – it’s just science and technology.

      Oh, and I am a religious person – so please no pro-religeon responses- I was simply saying that religion is up to debate – not everyone BELIEVES the same thing- but that with math and science we KNOW … know vs believe, thats all.

  • Just a quibble but aren’t most the advantages he mentions due more to the RAW format than the 12-bit space?

    Does a 12-bit ProRes 444 file have anywhere near as much flexibility as we see in this example?

  • The BMCC is a dream come true, but camera operators will have some nightmares.

  • My RED Scarlet definitely can’t recover anywhere near that much detail in the highlights, it is 12bit as well. Some fanboys even argue it’s 16bit. I’m pretty sure not even the 5d Mark III can do this in a RAW Photo. But isn’t each frame on the bmcc in dng mode about 5mb, holy HDD overkill :-) Can’t wait to get my hands on one though

    • The Scarlet has a smidgen more dynamic range than the BMCC, so if you’re having trouble with detail in highlights it’s probably because of how the dynamic range is being spread between highlights and shadows. If you need more range in the highlights you should expose for a higher ISO value. If you need more range in the shadows you should expose for a lower ISO value. (This is also how the Alexa works incidentally)

      Usually you can get everything exposed so that you’re not clipping in either shadows or highlights…just look at your histogram and make sure you’re not getting red bars rising on either side…a red bar on the right indicates clipping in the highlights, and a red bar on the left indicates clipping in the shadows.

      Also make sure that you’re not clipping something in post. Working in a tool with 32-bit floating point precision is going to give you the best results.

      • Re: the Scarlet having more dynamic range… Have you taken any measurements yourself? Whose numbers are you trusting?

        • Pro Video Coalition’s test of the Epic vs Alexa seems to indicate between 13 and 14 stops of DR for the Epic, so I figure 13.5 is a reasonable average (and one that seems to be commonly quoted for the MX sensor). The BMCC is said to have 13 stops. It would be awesome to see a direct comparison though because DR can be a difficult thing to objectively pin down.

      • Thank you, I know how to expose my camera and like I said: What we see in the Video above it not possible with a Scarlet. If you expose to keep all the highlights in a comparable scene you end up with unusable shadow detail.
        There’s the hdrx option but it comes with a lot post and I really haven’t seen any hdrx video that looked that much better than a shadow/highlighted spot on one exposure.

  • This article is pure CR*P.
    Sorry to be so hard. The results are misleading because the methodology in this so called test is simply wrong. It doesn’t have anything to do with 8bit vs 12bit.

    The mistake is in the conversion that he makes from raw to bitmap. Again, nothing to do with the color depth. (of course 12bit is going to be better than 8 bit, but in a completely unrelated topic to the one discussed here) When he is using Adobe camera raw to convert the raw data into a bitmap, he is clipping values in the highlights (not a mistake per se) and so the problem comes when he comes back and compares the native raw file to a clipped bitmap. He can recover highlight details from the raw because he is still accessing the whole data captured by the sensor. If he had done the same conversion to 10bit or even 12 bit uncompressed as opposed to 8bit, the same issue would appear.

    It’s your workflow dude what’s wrong and so you are getting the wrong interpretation!

    • So how else could he do the test with the same image, without having to shoot it with two different cameras? It can’t be done. If the test had been done between a BMC image and a 5DIII image (or any other 8-bit camera), it would open up a whole host of variables (and complaints).

      But if it is so wrong as you suggest, I’m interested to hear your perspective on how it could have been done better.

      • Well, there are just too many variables here. What is the point of the test? To compare how an 8bit vs 12bit image behaves under heavy color correction?

        If that’s the point, which is not exclusive to a camera manufacture, just have the same exact frame exported at those two color depths and THEN compare.

        What he seems to be missing is the fact that adobe camera raw is a conversion tool that uses the whole linear raw data of the camera to try to make a pretty image. Many people think that raw is some sort of holy grail in which you can magically recover lost highlights. This misunderstanding comes from how adobe camera raw works (and similar applications), by having a certain amount of contrast and a gamma curve already applied when you first open the image. Then if you lower the contrast or use the highlight recovery, you seem to be getting extra detail, but in truth, it was the software who was hiding that detail in the first place (sensors capture light in a linear way, so once the image is clipped it’s clipped.)

        Ideally when doing the conversion raw-image you want to preserver all the information that the sensor was able to capture. Some professional cameras do this in camera (Sony Slog), some let you do it in post (Red). You’ll notice that with R3Ds you can’t get any highlights back once they’re clipped like these guy does with the DNGs from the Blackmagic. That’s not because the blackmagic is awesome, but simply the way the workflow works. RedCine will just show you the whole dynamic range if you load the right tone curve. Again, there is no magic here.

        Again, if you want to compare 8bit vs 12 bit, an easy way would be to export the same frame from a Red clip as 8 bit and 12 bit and then do the grading to see the difference (8bit will show some posterization under heavy work, but that’s it). If you want to do it with the Blackmagic camera, make sure you preserve all the data when exporting from adobe camera raw into an image file (I suspect lowering the contrast until the histogram is not clipped anymore would work).

        Still, the whole comparison is kind of a waste if the goal is to compare 2 cameras…

        • You do make a good point Mac.

          I think perhaps the article/video name was simply a misnomer. Perhaps it should have been about emphasizing the difference between RAW and baked-in bitmap images. There is a clear advantage to RAW there.

          I agree when he exported the RAW file he should have used a flat/linear profile to get way more DR from it, though the fact that the original footage was 12-bit and can be adjusted in post to create a superior 8 or 10-bit image… actually no, that’s still just an advantage or RAW, lol.

      • I really don’t understand what Mac is talking about. I am just wondering if Mac could kindly bother to display his point of view with examples in a video.

        • Mac did a lot to explain what was done wrong – he tool the time to write a LOT about it – and in two responses!

          I am having trouble understanding what it is that you can’t understand about it… it’s pretty straight forward. He took the time to explain what was wrong – if you don’t understand it why don’t you google it all, item by item and figure it out.

          Why do people think that just because you prove someone wrong about something that you somehow owe people even more effort to show them an example. Maybe he will make a video, but are you paying him for his time? Are you his boss? why can’t people do their own work for once?

  • 1080p vs. 2.5K? By it’s very definition, you’ll get more detail, guys. I think you missed that.

  • even if you have a lousy or good story, audiences won’t care what camera you used–

    • I was just watching a movie called “Ben Banks” last night. While most of the film looked “okay” there were some scenes with extensive Moirée in the background. Maybe some people do not notice it or aren’t bothered by it. I for one thought it was horrible. It even distracted me so much that I didn’t pay much attention to the actors anymore (in that particular scene that is). Maybe nofilmschool has made me more aware of these things :-)

    • I forgot to mention, the movie was filmed on 5D Mark II’s

      • Lol.. This is the thing. If i wasn’t into filmmaking, i wouldn’t be looking for that stuff.. I saw a comparison screengrab between the gh2 and the gh3 the other day.. It was a lamp… I’m like.. “Yep…that’s a lamp.”
        It’s only until i scroll down do i see people going apeshit. “Banding!!!!!!” I didn’t know what the hell i was looking for.. I know what it looks like & all.. But… Damn. In our struggle to find a camera that we can afford with righteous features, it’s almost taken over. After i saw “Genesis”, i scrolled to the comments.. “Noise @ 0:53!!!” ..”Banding in that corner @0:26!!!” I mean, are they going frame by frame? Then it’s “f–k this camera.”
        So now, there’s a cinema camera that plugs out raw 2.5k images at 12 bit with 13 stops of DR for $3000… There really should be no excuse to get out there and make your movies now.. It’s funny because now that these kinds of images can be had affordably, this is where we’ll start seeing the community weed itself out from the youtube cat videos.. Someone will say, “the sensor is too small, so i’m just gonna wait… F–k bmc.”

  • I am a fan of everything about the BMCC except the sensor size/mount combination. When/if Black Magic releases a APS-H or even APS-C camera with shallow ffd, I will be the first one to sell my Canon 7D (soon to be EOS-M).

    I will still keep my 5D MKIII for its full-frame sensor. But, if BMD releases a full frame camera (or even a 4:3 sensor with anamorphic capabilities) sub-24mm FFD, I’d burn my 5D in effigy and go 100% BMCC.

    I’m also a big fan of them using the CinemaDNG format.

    If they had an APS-H I would then stay away from Red and begrudginly Arri until they can decide to move from the proprietary dark side.

  • Marco’s 8bit vs 12 bit was a definitive comparison hands down. On another note: Why does everyone keep saying the BMCC is only a $3K camera? It doesn’t come with any lenses. If you add in 3 basic primes at $2K each, throw in a better battery, SSDs, and some accessories at $2K… now your $3k camera is really an $11K camera. And if you plan on shooting even a quality short or small feature, you will need a minimum of two cameras so now your minimum starting is $22K. Am I too conservative or overshooting the money math?

    • Joe Marine on 11.1.12 @ 1:36PM

      Right, but you don’t need 3 $2,000 lenses, and many people have some lenses already. It’s still a $3,000 camera, and it’s completely up to you to outfit it the way you want, and you can certainly spend as little or as much as you want.

      • You are perfectly right. I have an old Leica R lens, a couple of Nikorrs and a old cannon. The only one lens I’d need is an ultra wide for my BMDCC. People add all sorts of mathematical calculations to the cost of BMDCC forgetting that these extra costs need to be included in all the good cameras like Arri, Red, Epic, Scarlet and so on. So what’s the problem. BMDCC is still relatively cheaper,

    • Daniel Mimura on 11.6.12 @ 5:07PM

      A car price does not include gas. What’s your point about add-ons?

      Every camera requires add-ons. A tripod would be the first one.

      • Daniel Mimura on 11.6.12 @ 5:12PM

        I already have v-mounts, Bartech Focus Device, power cables for p-tap or 4pin xlr for power (assuming it’s the same size as for their hyperdeck shuttle power), matte box, lenses (including 11-16), tripod, steadicam, SSDs. So yeah…$3k.

    • In that case – you wouldn’t be able to quote any camera pricing -

      Oh, whats that- you want to use the line up of 8 arri master anamorphic lenses at 43,000 each, well then, this BMC camera will cost you 347,000

      Oh, whats that you say, now this other user wants to use rokinons – well, thats a deal because then the BMC camera will only cost you 4,000 dollars. LOL

      Should the price of the brain only be counted if it needs an external recorder or something? NO
      should you include the prices of items you will use on any camera? NO

      Whats next – you can’t make a movie without lighting so add on the cost of lights to the cost of the camera
      does the camera then cost less if the director uses only available light?
      OH NO- what if he uses the camera with available light on one project andl lights on the other – do you say the camera cost more on the one with the lights and less on the other? LOL

  • Gary Simmons on 11.1.12 @ 9:38PM

    I think its premature to comment till we see if the mark III update from canon does with the HDMI Output as promised then some magic lantern tweaks on top of that before we can decide what the ultimate DSLR quality will be. I started with a DSLR because I already have a investment in canon lens because I started as a still shooter and as I have learned from the good folks on this and other sites I get better performance out of the same gear. I recently got to borrow a L series canon glass and wow what a difference I learned 2 things I need better glass and my images are better than I thought as I have tweaked the camera past what it could do with standard glass. So no matter what you have, learn how to work with it and you will get better results.

  • Erik Stenbakken on 11.2.12 @ 12:31AM

    Great stuff. Thanks for the article and video. Lots to absorb and consider here. There are likely a whole host of us who are shooting on DSLRs and doing ok. But if others are where we are — we’re looking for the next step, and this is valuable information. Thanks!

  • i am sorry, but this is a wrong testing method.

    i am not saying 12bit footage doesnt offer greater flexibility than 8bit.
    ofcourse it does

    but to compare 12bit footage to itself, after its converted to 8bit using the default settings of camera raw software? the results are as expected. lots of information has been thrown away during this conversion.

    on the other hand, i am not sure other cameras throw away that much information during the internal conversion to 8bit.

    • The 12-bit converted to 8-bit still retains a higher level of quality than if it had been shot from an 8-bit initially. You still have color information retained converting down than where you possibly had no color information (clipped or crushed) to begin with from the 8-bit. Mathematically (8-bit = 256 values and 12-bit = 4,096 values) 12-bit kicks the dog snot (even converted down) out of 8-bit… end of story.

  • Erik Stenbakken on 11.2.12 @ 12:37PM

    Hey, if readers find fault with this methodology… look at the FIRST test he did. Remember that? And the crappy quality of the DSLR? (the very DSLR I use). Folks bitched and moaned that THAT test wasn’t fair for this and that reason. Unfair comparison, many said. Settings were off on one, etc. (hey, I said that to myself and others in-company). Then the test was re-shot using the SAME camera and … we bitch and moan. Could it be, just maybe, that with the right workflow the BMCC* is just kicking my camera’s chip up and down the street? (*or any 12 bit camera, for that matter).

    Yeah, I didn’t want the workflow of RAW when shooting stills either. “Jpeg is just fine!!!” … until I found the power in RAW – and I might add: an efficient workflow that allowed its use.

    If we didn’t like the results with two different cameras; and we don’t like the results with one camera; maybe it’s just that we don’t like the results (and what they are saying).

  • Hi guys. Do you know where i can buy a Blackmagic Cinema camera? Thanks.

  • “As far as dynamic range is concerned, there is only so much you can do with 8-bit to protect the highlights.”

    That is not entirely true and it is extremely important when comparing file standards of the same bit-depth or even different ones to understand the very clear distinction between linear, log, gamma, and REC709 video standards – all of which can either increase or decrease the apparent overall dynamic range stored in the digital signal somewhat regardless of bit-depth.

    For example storing linear images requires very high bit-depth since after gamma correction is applied the blacks are stretched apart and less data information is available to describe the dark tones. At the same time too much data is used to represent the bright tones. This is just how linear works once we process the image to the right gamma needed to view it.

    The Alexa is a good example using a 10 bit Log signal to carry almost as much actual information as a 12bit linear signal does, and a 12 bit Log file can carry almost as much info as a 14bit linear signal. The Sony F3 can shoot SLOG in 8-bit or 10-bit and because of that you will be able to capture higher than normal apparent dynamic range than if you were say recording in the REC709 standard which by its very design throws away dynamic range due to its original intended purpose of display images on displays with limited “dynamic range”.

    In the end, it’s not as simple as looking at pure bit-depths for comparison.

    • Sony’s latest DSLR the A7s records in only 8 bit but has the option to record in SLOG. How does 8 bit SLOG compare to the 12 bit in the black magic cinema camera or the very new kinemini?

Comments pages: 1 [2]