Against High Resolution Large Sensors – Why I prefer to shoot in 2K Super 16

Following the official Blackmagic Forum and unofficial BMCuser forum, I feel that there is to much attention being made on the Blackmagic URSA line of cameras. My qualified guess is that there are a lot more Blackmagic camera users that own and shoot with a Blackmagic Cinema Camera (BMCC), a Blackmagic Pocket Cinema Camera (BMPCC), or a Blackmagic Micro Cinema Camera (BMMCC). So why is there so little said or discussed regarding the BMD Cinema Cameras? Why all of this unequal focusing on 4K (and above) resolution and Super 35, at the expanse of 2.5K (and below) and Super 16? In my opinion, it’s time to bring back (and reclaim) the topic of Super 16 CinemaDNG RAW shooting; what started it all. I still haven’t given up on the original Blackmagic Design mission of bringing forth and establishing the small form factor format, making CinemaDNG RAW accessible to the masses and guerilla filmmakers through the Super 16 format. The democratisation of filmmaking. I am making a case for the original Blackmagic Cinema Camera (BMCC), which I consider to be a oversized Super 16 cinema camera, rather than in between Super 16 and Super 35, as well for the dedicated Super 16 Blackmagic Pocket Cinema Camera (BMPCC), and also the newest addition, the Blackmagic Micro Cinema Camera (BMMCC). I am quite daunted by all of this talk about the URSA and URSA Mini, which (based on discussions on these two fora) seems to be the Emperor’s new cloths (read, product in the making). This also applies to the URSAs predecessor, the Blackmagic Production Camera 4K, which sacrificed dynamic range for a global shutter; the first URSAs were fitted with that same flawed Super 35 4K sensor. However, what can be brought against the URSA may also be brought against other brands that shoot with Super 35 sensors (or larger) and 4K (or more), such as the RED, Canon, Sony, and Panasonic cameras.

Blackmagic URSA Mini Pro – The Emperors New Cloths?

Seems to me, reading various fora and threads, that the BMPCC is still the most reliable, affordable (as in bang-for-the-buck) BMD camera and also the one most used in the industry. Since 19 March 2015, with the Blackmagic Camera 2.1 update, it has reached its full potential and come to its age; it’s not the same camera as in 2013, where most reviews on the Net stem from, making them outdated today. The BMMCC also seems to be equally reliable, drawing on the development of the BMPCC. There are some good power solutions, such as the JuiceBox, with reports stating it can make the BMPCC shoot RAW over 7.5 hours straight. That’s almost unbelievable for such as “small” battery; I do like its size and prefer it over a “V” mount, suiting the small form factor of the BMPCC better. I have only used SanDisk Extreme 64 Gb SDXCs so far, however the BMPCC also accepts 512 Gb cards for much longer recording times; this is really a game changer, in combination with the JuiceBox Magic Power! However, there doesn’t seem to be any means to work around the Micro HDMI connection, the Achilles’ heel of the BMPCC, which we have to live with. This makes a case for the BMMCC. (I’m almost regretting that I didn’t buy it instead of the BMPCC.) On the other hand, the strongest point of the Pocket is its ability to shoot straight out from the box, after applying a lens to it. On my latest short, The XmasTree Hunt, I held it as a stills camera, in my right hand grasping the grip of the camera body; I really enjoyed holding it in this way througout several hours of shooting, without any strain and walking through foliage. No accessories attached; I could easily sneak in to the action without creating unecessary attention. It is truly a guerilla style camera. Also, it has contributed tremedously to zero budget filmmaking throughout the world. It is also, amongst all BMD cameras, the choice of several rental houses and some big productions, such as A Tale of Love and Darkness, and Avengers: The Age of Ultron. Both amateurs and professionals love it. It deserves our respect. And I am proud to own one.

Blackmagic Pocket Cinema Camera – A most successful design

The URSA family of cameras, that’s an entirely different story (although it seems to have overcome some of these shortcomings with the recent introduction of the URSA Mini Pro). Also, it seems that the classic Blackmagic Design colour science (that we all have come to love) has deviated since the introduction of the BMPC 4K and the URSA, contrasted to the Blackmagic Cinema Cameras (i.e. BMCC, BMPCC and BMMCC). Even if we do not take into consideration distracting image quality issues, such as fixed pattern noice, cross hatching and magenta tinting, I in particular find the URSA 4.6K sensor to present to sharp of a picture, taking the filmic sensation away that is present in the smaller sensors. Personally, I would had preferred to see Blackmagic Design invest in and further develop its classic models, the BMCC and BMPCC (and BMMCC), the Super 16 format and sub 3K standard. And not as now, only focusing on the URSA Mini in the firmware upgrades. That’s what Blackmagic Design have excelled in and deserves most credit (and have gained the best reputation) for. In this post I will explain why I prefer to shoot in Super 16 1080p HD and why I ignore the 4K+ hysteria. I will also explain why I still prefer to shoot in the Cinema DNG RAW format rather than ProRes or DNxHD/HR. Why I prefer Digital Film over digital video. I will address both the pros and cons of Super 16 CinemaDNG RAW filmmaking in general, and Blackmagic Design Super 16 cinema cameras in particular. However, what can be said concerning the BMCC, BMPCC and BMMCC also spills over to the other two classic CinemaDNG cameras, the Ikonoskop A-Cam dII and Digital Bolex D16 (feeling a common bond in the brotherhood of Super 16 CinemaDNG filmmaking).

Ikonoskop A-Cam dII – The original CinemaDNG Digital Super 16 Film Camera

Are there any cons with the lower resolution 2K small size Super 16 format? Of course there is. Digital Super 16 Film cameras behave as emulsion film cameras, in that they require lots of light to get your desired picture. Super 16 cameras, such as the BMPCC, create lots of challanges with tungsten lighting. This, in combination with the limited storage of the SDXC (although that may be worked around it seems), makes shooting with the BMPCC and BMMCC very filmic, not only in the final result but also in its operation. On the other hand, this is perhaps one other aspect of what is so appealing with the format, that the BMPCC behaves as an old Super 16 film camera in its workflow. However, as I am not proposing the total abandonment of Super 35 sensors with 4K or higher resolutions, there must be some pros in using large sensors and high resolution. Even I can imagine that 4K or 4.6K has its place in very wide shots of landscapes or cityscapes, and such b-roll footage, which perhaps may become a little to soft and with to much lost detail in a BMPCC or BMMCC. There may also be some benefits in using the 4K+ Super 35 format when shooting miniature visual special effects. But here I would probably rent a camera for these types of shots if needed for a larger project, and stick to my Super 16 format for the narrative. Personally, I would probably settle for a BMCC for extreme wide shots of landscapes, or minitures, as it blends so well with the BMPCC and presents 50% more resolution. I’m more a friend of intimate cinematography. However, when it comes to digital projection, 4K is definitely the way to go, in particular when showing old celluloid classics. When it comes to the resolution of photochemical celluloid emulsion film, I once did some research on the Net of the equivalent digital resolution natively, which is always an approximation as celluloid film resolution is based on the size of chemical grain not pixels, and grain not only lays itself haphazardly but varies in size within the same frame as well. However, I came to the following conclusion:

8 mm film = 2K (1-5 Megapixels)
16 mm = approx. 4K (10 Megapixels)
35 mm = circa 6K (20 Megapixels)
65/70 mm = approx. 8K (30-40 Megapixels)

The pros of shooting with 2K Super 16 are several. This is partly a question of using managable file sizes. If you shoot Super 16 in 1080p or 2.5K and lossless CinemaDNG, you will have more managable files compared to 4.6K lossless CinemaDNG. With 4K or 4.6K, you will have to compress the RAW capture “lossy” to get down the CinemaDNG file size. I rather “sacrifice” resolution over compressed RAW; I want nothing less than uncompressed or losslessly compressed RAW. Unfortunately, Blackmagic Design no longer offers uncompressed CinemaDNG RAW in any of its cameras. Reading Kurt Lancaster’s Cinema Raw: Shooting and Color Grading with the Ikonoskop, Digital Bolex and Blackmagic Cinema Cameras (2014, Focal Press, New York, USA), from 2014, he states that the BMCC shoots uncompressed CinemaDNG RAW. However, the Blackmagic Design web page only claims Lossless Compressed CinemaDNG RAW. This is due to uncompressed RAW being removed in a later firmware (version 2.1), where it was substituted with lossless compression. I cannot help but wondering, why BMD didn’t keep the original uncompressed RAW as an option, to choose between losslessly compressed and uncompressed? I know that many will say that there is virtually no difference between uncompressed and losslessly compressed RAW.¹ However, BMD themselves claim zero loss of detail but only mild compression with about 35% file size reduction for very blurry / underexposed / overexposed images ranging to around 25% file size reduction for well focused, well exposed, detailed images, which is the equivalent of 1.5:1 compression. Now, 25-35% of removed information is quite substantial, although there is no difference to image quality and colour depth, etc. This should be the choice of the cinematographer, not the manufacturer. The Blackmagic Pocket Cinema Camera has had lossless compressed CinemaDNG RAW from the very start, and here it is fully understandable as it records to small SDXC cards. But the BMCC records to SSD and should be able to handle uncompressed RAW. Its obvious that Blackmagic Design have moved more and more into compressing their CinemaDNG files because the industry requires it, disregarding the independent filmmaker and his need for creative control. Today, even the BMMCC offers lossy compression of CinemaDNG up to 3:1, which is redunant in my opinion (if not shooting 60fps); with 4K or 4.6K cameras (who also offer 4:1 compression) this is almost a necessity to be able to manage files.

Blackmagic Cinema Camera – The Firstborn

However, the argument against this last premise is that some claim that one may downscale 4.6K or 4K RAW to 1080p, in an effort to reduce the file size; this counterargument claims that the image will look better as there there will be better color sampling, much less aliasing, and a finer noise pattern. Some even claim that 4.6K RAW lossy compressed down to a size similar to the BMPCC will look as good (but with less aliasing) when delivered at 1080p, as the files contain approximately the same amount of color information; perceptual based compression is claimed to attenuate high frequencies similar to a low pass filter, which is exactly what is desired to do before downscaling images from high resolution to a lower resolution. However, this is the school that goes contrary to the less (compression) is more philosophy of the Super 16 2K proponents. These will probably never be recociled. As I belong to the latter, I allow myself to be highly sceptical of this last thesis. And these shools will probably never convince each other until they meet in a 2K theatre, looking at the same scene (or film) shot on a BMPCC (or BMMCC) mounted parallell with the URSA, fed directly from Resolve in RAW (or at least in DCP/JPG2000). That said, Iv’e been pondering the counter argument a lot and come to the conclusion that this also applies to a sensor with a certain amount of pixels (say 2K) but perhaps four times as much photosites, i.e. four photosites per pixel, (which would amount to 4K of photosites). So, each pixel contains one blue, one red and two green filtered photosites (i.e. classical Bayer pattern). But still records natively in 2K but with larger colour depth in bits compared to 1:1 photosite-pixel ratio. This means that a “pixel” in reality is a digitized frame containing four quarters of physical photosites, each with its own colour (with the exeption of the green photosits which are repeated), which are unified in the downscaling process into a whole to represent more colour information. For me, this concept is easier to grasp when I think of photosites vs. pixels. So when we talk of a 4K Super 16 sensor, we actually are referring to a sensor with 4K of photosites, which are downsampled / downscaled to 2K or 1080p of digital pixels in the resulting image.

The BMPCC Super 16 2 Megapixels sensor

So I might accept the premise that 4K downscaled to 2K or 1080p might create a better colour resolution and higher virutal (or “simulated”) colour bit depth in the process, creating more colour space but retaining a natural softness to the image, to enhance the filmic photochemical feeling.² But only with CinemaDNG RAW and provided that it is at least losslessly compressed, i.e. only if downscaling isn’t affecting compression, other than reducing file size because of lessening the resolution; colour space compression is a big no-no. Because, somewhere inside of me it nags, knowing that something is “lossy” compressed, with big chunks of information removed, although officially named “CinemaDNG RAW”. Something ain’t right here. One has to remember that CinemaDNGs are RAW data pulled straight from the sensor, and if you start scaling or compressing things then it isn’t really RAW data from the sensor anymore. So, if you want to work with a 2K RAW workflow, the only viable solution is to shoot with a 2K sensor. Another important aspect to consider is that larger photosites are able to collect more photons in their “buckets”, resulting in a higher Dynamic Range. Lower resolution sensors have larger photosite pixels. So, in simple theory a 4K Super 16 size sensor wouldn’t be able to capture as much Dynamic Range as a 1080p or 2K Super 16 sensor. What complicates things is that not only “bucket” size is important but “bucket” depth as well. So, it seems that both width and depth is important to create dynamic range. Comparing pixels with equal depth but different width would yield different dynamic range, as well as two pixels of the same width and different depths. So, in conclusion, the sensors yielding most theoretical dynamic range are those that have both wide and deep pixels or photosites. This still gives an advantage to 1080p or 2K in Super 16 sensors. Just make their pixels deeper to collect even more light! While many call for a 4K Super 16 sensor, I personally wish for a 2K sensor with a deeper well or better dual-gain architecture, to squeeze out more dynamic range than the current 13 stops. So I guess that there is a trade-off between Dynamic Range and Resolution (or if downscaled, Colour Resolution). Do you prefer Resolution over Dynamic Range? Do you prefer detail over exposure control? Not me. So we’re back on square one.

Blackmagic Micro Cinema Camera – The Last in Line?

A much more important aspect than mere technicalities is the aestetical one, however. It is from this artistic choise of the filmmaker that the school that I belong to was formed. I didn’t invent it, and I am probably its worst proponent (being a technical ignorant). The guys and gals who designed the Ikonoskop A-Cam dII and the Digital Bolex did, as well as cinematographers such as Michael Plescia, Kurt Lancaster, et al. As with statistics, you may prove anything with mathematics. It finally comes down to how you subjectively perceive the image. When looking at the URSA, I think “nice image”. But I still prefer the ALEXA with its 3.2K Super 35 sensor, that downscales to 2.8K ARRIRAW, anytime over the 4.6K of the URSA. The fact is that I find the image coming from the BMCC, BMPCC and BMMCC speaking to me more than that of the URSA. I prefer the colour science of these cameras over that of the URSA, however much better the latter’s is on paper. Many professional cinematographers claim that the original BMD Cinema Cameras match the ALEXA perfectly, both in colour and in sharpness. As with the ALEXA, I prefer the somewhat softer or diffused image of faces coming from a 1080p or 2.5K sensor over the hyperdetailed and ugly-making 4.6K sensor. Here I agree with Michael Plescia’s and Kurt Lancaster’s arguments against higher resolution above 3K (see Cinema RAW). This is the same urge that drives celluloid filmmakers to prefer the Super 16 image over the Super 35 film stock image (which starts to look “digital” in their minds). Then there is the issues of small form factor, cheaper cameras and vintage 16 mm or Super 16 lenses (that create as wonderful images as those of Super 35 for much less price). I own two 16mm cameras, the Krasnogorsk-3 and Kinor-16SX-2M. Both are regular 16. I will never part from them. I have scrambled their glass for my BMPCC, so in a way these cameras live on through my Pocket. That how I see the BMPCC, as a Digital Super 16 Film Camera. Not as digital video; it behaves as film.

Arri ALEXA – The Holy Grail of Digital Cinema

Resolution isn’t everything, and after a certain point isn’t the most essential aspect of a good image. Regardless if the theoretical colour depth increases with each increased photosite or pixel, the colour science of the camera and curves seems to be of more import. At least to my higly unscientific eye. As much as I hate to admit it, as I really like and respect Blackmagic Design for all of their efforts to bring the big cinema to the people, I much prefer the general look coming from the 2K Digital Bolex over the URSA 4.6 or any of the RED monsters; it almost emulates the image of the Arri ALEXA 3.2K. If I had the money, I would prefer to buy the Digital Bolex D16 over a Blackmagic URSA Mini Pro 4.6K anytime, not only because of the cheaper price tag. The only current Blackmagic Design camera that has caught my interest is the BMMCC, because of its 60 fps and professed better sensor refresh rate, handling the rolling shutter effect better. I might buy one in the future. As the colour science of the BMPCC and BMMCC is great as well, it would probably be possible to push and pull the images to emulate the filmic look of the D16. I have not forsaken Blackmagic Design. I still love my Pocket Cinema Camera, and its organic image. I’m just saying this for the sake of an argument, that the subjective experience of any image is much more important than maths and numbers. My eyes, and probably most other’s, loose the ability to se any difference post 3K. But a colour scientist (or rather artist) may create wonders with the RAW sensor information coming uncompressed and unscaled from such a small sensor as 1080p or 2K, that I and others will be able to enjoy with much pleasure. For me, this is movie magic and cinema mysticism.

Digital Bolex D16 – A Hipster’s wet dream

I guess that the RAW 2K Super 16 format is a particular niche that attracts relatively few enthusiasts, compared to the larger following of the 4K+ Super 35 shooters. That’s probably why Digital Bolex couldn’t survive in this age of resolution hysteria, plus it being too expensive for the independent filmmaker. (I like to believe that the BMPCC has been one of BMD’s most successful camera designs when it comes to sales, as it was the cheapest camera delivering RAW footage, not that it was a Super 16 camera per se; however, I suspect that many previous BMPCC owners have gone over to buying 4K cameras with larger sensors, even DSLR ones.) Although the Digital Bolex D16 is affordable, as in cost effective, as in cheap in relation to what it offers (especially when compared to the Ikonoskop A-Cam dII which was way overprized), it was beyond my budget which had a 1K limit; if the BMPCC (or the BMMCC) hadn’t existed, I would probably had to settle for one of those smaller 1080p Canon cameras and hacked it with Magical Lantern to get RAW, or perhaps for the GH4 with V-Log L. It’s obvious that we who prefer to shoot RAW 2K Super 16 footage see something in our images that others don’t or even dislike. I believe that the closest thing in describing this feeling would be movie magic. I believe this small format provides more “magic” and “mystery” to our images. Many of us who belong to this category, love old looking minimalistic lit black and white films. Some of us even prefer looking at old German expressionist silent movies, over more recent remakes or adaptations, because of the image quality. It’s not literal, it’s subjective, and therefore emotional, and profoundly human. I suppose, we belong to that same category who prefer to listen to crackling LPs over CDs, or reminiscing with nostaliga how we once experienced the magic of movies looking at our old VHS tapes, compared to the clinical Blu-rays of today (some of those cheesy looking films of the 80’s look better on VHS than on Blu-ray, as that terrible low res footage hides the blemishes).

Silicon Imaging SI-2K Digital 16mm Cinema Camera – The Predecessor

As an alternative anology (and diagnosis of my mindset) I still watch my DVDs and Blu-rays on a over a decade old DLP projector, which has a native resolution of 1024 x 576, i.e. 16:9 standard definition. It may recognise upscaled HDMI signals up to 720p and 1080i; I prefer to upscale my DVDs to 720p and downscale my Blu-rays to 720p, as I noticed that my projector downscaled 1080i to 540p; there is lots less aliasing in 720p compared to 1080i. And I still haven’t upgraded it to one of the newer HD projectors with all the latest bells and whistles, as I (and the rest of my family, and friends) am fully satisfied with the image that we see on the white screen, which involves me in the story and the vision. The particular model has a reputedly good colour science which I feel reluctant to trade-off for higher resolution, which I could get for half of the price today that I spent on it 12 years ago. I did a major shootout with my projector, just after that I had bought it, comparing it to much more expensive and brand new HD projectors (my projector already was an older and discontinued model when I bought it with a discount), and it did very well impressing many present. It has a very sharp Zeiss lens, so I even have the focus off a bit to soften the image and take away some of the low res screen-door grid pattern. I have no plans of buying a new one, as I love the image. In my belief system, it is good to be content with what one has, as there is a tendency in the film business, both at the producing end and the receiving end, to get the latest gimmick. It is good to take a deep breath, calm down and actually reconcile oneself with what one has, and take pleasure from it. If I had the money, I would still shoot on Super 16, using a crystal sync modified version of the old Soviet manufactured Kinor-16SX-2M. That’s how old school I am.

Kinor-16SX-2M 16mm camera with LOMO 16 OPF 12-1 f/2.5 10-100 mm

We all love the classic look of celluloid film, although it is far from high fidelity. With classic I’m referring to older film stocks up to the 1980’s. We even prefer the original Super 16 film format over Super 35 film, as it captures the emotional content and movie magic better than modern Super 35 film stock. The big digital camera manufacurers expected us to be extinct. But we’re still here. And the Blackmagic Cinema Cameras (i.e. BMCC, BMPCC and BMMCC) are the closest thing (still in production) to offer us a viable alternative to our beloved celluloid dreamings. We want our cameras to emulate that photochemical feeling. I suppose it has something to do with conditioning; perhaps many (or most) of us grew up in the 1970’s or 1980’s and learnt to like the look of our favourite movies. And today we want to capture that same feeling and mood, or even recreate or reenact the movies of the 70’s and 80’s. The point I’m trying to convey here is that mathematics, statisctics and diagrams won’t persuade us to leave this format (i.e. RAW 2K Super 16). We prefer the art of film over the science. We aren’t looking for the perfect image, that captures reality objectively, analytically and clinical. That is for the Super 35 6K cameras to capture. No, we look for distorted reality. We want to escape reality. We want to enter the reality of dreams, of magic, of myth. We are dreamers, escapists, romantics. Hoplessly fallen, beyond redemption. Ikonoskop and Digital Bolex tried to reach out to us, and believed in us, as they came from the same stock of humanity as ourselves. Blackmagic Design also reached out to us for a while, at least. But their passion at heart lies elsewhere, I think; they want to play with the big boys. Although the introduction of the BMMCC proved that they (BMD) at least have retained a sincere respect for us. But it was a pipe dream, doomed to fail. That’s why companies such as Ikonoskop and Digital Bolex eventually withered away, bringing forth niche cameras, suited for small production runs. Not to much of a profit, I guess. Investors don’t like that. They don’t like us. BMD is smart enough to feed both spectrums of the market, and hopefully they will continue to provide us with our dreams.

¹ BMCuser forum member “cpc” states the following: “There is zero difference between uncompressed and losslessly compressed. ‘Zero difference’ is not the same as ‘virtually no difference’. Zero difference means no difference whatsoever after the compressed image is decompressed. Exactly the same pixels. Looks like you are confusing information with data, which leads to your conceptual problems with compression. The purpose of lossless compression is more efficient representation of information using less data, without losing any information. This is possible because the uncompressed representation of information is not optimal in terms of binary encoding. When you zip or rar something it comes out exactly the same as the original, once you decompress it. Same with lossless compression for images.”

² BMCuser forum member Alex Michell states: “You can’t generate more precision (10-bit) from less precision (8-bit) – all you can do is generate intermediate values by way of dithering during scaling. Make no mistake; the new 2K file generated from an 8-bit, 4K source will have more than 256 values encoded in it but none of them will be derived from information gathered by the sensor.” I guess the same may be stated when downscaling a 12-bit RAW 4K image down to a “14-bit” RAW 2K; it is a cheating which may be compared to emulating 50% slow motion from 24fps images using optical flow, whith a noticable lower quality compared to true 60fps, and possible artifacts that aren’t there in 24fps.


One thought on “Against High Resolution Large Sensors – Why I prefer to shoot in 2K Super 16

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s