DSLRs, Democratic Technology and The Cost of Bokeh: Part 1

DSLR filmmaking has been much ballyhooed in the last year or so. Cameras like the Canon 5D Mark II and Canon 7D have been hailed as the lastest in a long line of “democratizing” motion picture technology — inexpensive cameras that produce cinematic, shallow depth of field images that seem to rival the look produced by cameras costing many times more.

There were two reasons I was didn’t jump on the DSLR filmmaking bandwagon from the start. For one thing, in the last year I have been working on a lot of other projects, none of which involved needing to worry about how to use a new camera (finishing up a documentary and two DVD releases, raising money for a feature, and writing a script). Another reason was, frankly, I was skeptical. I saw photos of filmmakers dressing the cameras like this:

Going "indie" with a DSLR.

Going "indie" with a DSLR.

If that’s what you had to do to get it to work, I wasn’t interested.

As I mentioned in my previous post, though, I did recently decide to experiment with these cameras. And my uninformed skepticism has developed into experience-grounded ambivalence.

After a few months of wrestling with the cameras (especially the Canon 7D), I find them as frustrating as they are inspiring. Yes, I love the way they handle light. I love the lens interchangeability. I love their form factor, (at least initially).

But, as has been well documented elsewhere, these cameras have serious issues. Focus can be a challenge with their small LCDs. They’re prone to the “jello” effect. They shoot on a codec that is a challenge to edit and even more challenging to color grade. And, most frustratingly, they have major issues with aliasing, particularly moire, which is often not even observable while shooting.

Sure, I’ve had busted takes with other cameras — under-exposed shots on film that didn’t come out, shots that were a little soft in HD, or whatever. But DSLRs are built (or not built) for movie-making in such a way that you can very conceivably shoot for a significant period of time only to later discover that all of your footage is unusable. Not “disappointing” — unusable. Or perhaps you find something like this acceptable. (Note: I did not shoot this.)

To put it bluntly, these cameras have more red flags than a month’s worth of World Cup games. They carry a lot of risk for any serious project.

Because there are some undeniably awesome uses for these cameras, though, I have educated myself — by reading, by watching, by shooting — to find ways of working around their many, many problems. And I’ve learned to produce some nice footage.

But many of the techniques I’ve used to mitigate the problems involve spending more money, making the cameras bigger, and so on.

Fixing the problems often means taking away the very properties that make these DSLR cameras so seductive for filmmaking in the first place.

So in this post and the next I want to deal honestly with the basic costs of DSLR filmmaking and to consider whether these costs are worth the benefits.

It’s possible some DSLR acolytes will disagree with, or even have their feathers ruffled, by my writings about this technology.

That’s fine. These are my opinions alone and no camera is right for every person, at every time, for every project. But I think that by now this blog has established my credentials as a champion of smaller, less expensive, and simpler technology for movie-making. If I’m being critical, it’s probably for good reason.

So, today’s post will begin to consider the “cost of bokeh”, since their shallow depth of field is often touted as the leading reason for using these cameras.

My next post will finish pricing out the camera and accessories. I’ll also offer some thoughts on the notion of this technology as a “democratizing” force.

But enough prelude. What do we need to shoot motion pictures effectively with a DSLR?

Let’s start by going with a Canon 7D since it sits in the middle of Canon’s DSLR line, with the T2i at the bottom and the 5D Mark II and 1D Mark IV at the top. The 7D averages around $1700. That sounds like a bargain when you put it next to a traditional prosumer camcorder like the Sony EX-1R ($6300) or the Panasonic HPX170 ($3995).

(By the way, if you want to consider the costs with a 5D Mark II, which has an even larger sensor, add about $800 to our totals.)

Then you need a lens. If you want to want to get that shallow DoF then you need a lens that opens wide. And since many people have had good experiences using Canon’s Image Stabilized lenses, which seem to reduce some of the jello effect, we’ll go with Canon’s 17-55 IS f/2.8 lens. It’s been well reviewed and costs about $1100.

Since we’re trying to do this inexpensively, we’re only going to use one lens. If you want to take advantage of the Canon’s interchangeability (with, say, a cool Tokina 11-16MM), those are additional costs.

Some DIY filmmakers looking to get by on the cheap blanch at paying $1100 for a lens, but that’s nothing compared to a cine lens. In fact, just because you spent $1100 on that 17-55 f/2.8 doesn’t mean it’ll necessarily look sharp on the big screen. Shane Hurlbut, ASC argues that the only lenses Canon makes that are sharp enough for big screen work are their L-series primes. (Expect to pay $1300 or so for each prime and only the longer range lenses have Image Stabilization.) But we’re going to trust others’ reviews of the 17-55, which say it’s one of the sharpest lenses Canon makes.

(As a side note, you could go with Nikon AI-series still lenses. They’re both cheaper and are said to be sharper. But in my experience, you’ll need to buy a good Fotodiox Pro adapter [$70 each] to use them effectively. Plus, when you want to use your Canon DSLR as a stills camera, you’ll have no autofocus or auto exposure control, so I’m leaving them out of the conversation for now.)

We’re doing good so far, but sound, as they say, is half the picture.

While, technically speaking, one may record sound with the Canon, its sound capabilities are far from what you’d get with a prosumer camcorder (e.g., no XLR inputs, no level control, etc.). There is lots of work on Vimeo featuring beautiful shallow-focus images of flowers and so on, much of it set to cool music. But if you want to make movies, you know, where people talk and stuff, you’ve got to upgrade your sound.

I’m not going to count the cost of XLR cables, microphones, etc. since you would need that stuff with a traditional camcorder. Instead, we’ll just look at adding an adequate sound recording device. A lot of people using the Canon for DSLR cinema use the Zoom H4n recorder. It’s about $280. (If I were buying, I’d spend the extra $250 and get the Marantz PMD661 because it’s easier to use. But that’s just me.) An alternative is to use something like a Beachtek or JuicedLink adapter, but I don’t like the idea of all my location sound hinging on a single mini plug going into something that was primarily designed as a stills camera.

With the Zoom recorder (or similar) remember, you’re shooting double system. As such, you’ll need to slate your shots and spend lots of time in post syncing up your slates. (Or, if you don’t, reading peoples’ lips.) A time-saving solution is PluralEyes, which syncs your double system footage for you. Your time is worth something; PluralEyes has valued it at $150.

What’s the tally so far? $3230.

Oh. But we’re ready to make movies, right?

Yes and no. We may have picture and sound, but it may not be useable yet. But we’ll save that discussion for the next post.

14 Responses to “DSLRs, Democratic Technology and The Cost of Bokeh: Part 1”

  1. Jonathan Poritsky Says:

    Thanks for entering this conversation, it’s one that people take for granted. It’s very odd how quickly DSLR shooting, and tapeless workflows in general, took off. Moreover, HDV lived a short life, though in my opinion it bests many of the hot cameras today. Sure, we can talk bitrates and GOPs an color space, but HDV fit a workflow that, uh, worked. And with the right camera to capture a nice image on it (Canon A1, basically) you could make beautiful images.

    But that damn bokeh. Please, pretty please, in one of your posts pronounce the shutter dead, because that is the real cost of bokeh and tapeless and sensors and doo-dads. The DSLRs, by their nature, sport a rolling shutter, which is what offers the “jelly” effect as you call it. Even the RED, that savior of Indie inspiration, doesn’t have a shutter. A shutter! RED’s success is causing a trend where folks just look the other way because they are so in love with the static imagery they are getting from these cameras.

    Sheesh, this frustrates me. Unless we demand more from the camera manufacturers, we will get nothing. If folks fawn all over a company that can’t deliver a shutterless camera on time, then we’re doomed. Any insight on this matter?

  2. Paul Says:

    @ Jon: Thanks for the reading. I can certainly appreciate how you, as an editor, miss tape. I miss having a tape archive too, which is why I’m looking more and more at LTO. (Had a post on it earlier this year.) Sure, an LTO system is pricey, but it’s nothing compared to what a DVCPro or HDCAM deck cost.

    As for the rolling (non) shutter… I hear you. I don’t see it as much because it’s rare for me to use handheld, whip-pans, and so on. But I know it’s a huge frustration for others. What concerns me is that even bumping these little cameras causes your footage to look like a minor tremor went off in your neighborhood. That’s why the Image Stabilized lenses are essential, IMHO. The Red can withstand the bumps, that’s for sure. The downside is, a “light” Red rig weighs 20 lbs.

  3. Cheryl Hess Says:

    Hi Paul! I think the main issue is that people assume that just because the technology is cheap that DSLR is democratizing just as they said about miniDV. I would say that DSLR because of the technological challenges, is not a beginner technology. If you told two people with very minimal production experience to go out and make a 5 minute documentary and you gave one of them a PD-150 and the other one a 7D, I would bet on the person with the PD-150 to come back with the better footage, but then again shallow focus flowers over classical music makes me want to throw up. Having said that, I just sold my PD-150 on ebay and bought a 2Ti, not as a main camera but as a tool to have. Why not? You can get some great images out of them. As far as the Frankenrig factor, in my experience there has always been some of that with all the prosumer cameras. And if you think about high budge narrative production, it’s not like they are shooting on a bare naked Panavision. Oh, and you have a crew of how many people? I just don’t understand why people keep expecting a $1,500 camera to be an $80,000 camera. As with any camera you get you have to understand its limitations. I, for one, will not get my panties in a knot if someone’s tie has a little moire.

    So bring on the DSLR.

    FYI, this is what sold me on them (shot on the 5D):


  4. Paul Says:

    @ Cheryl: I’m in big agreement with what you have to say — I address some of this stuff with the second post (already written, scheduled for Thursday or Friday… can’t remember).

    The one thing I differ with you on is the moiré issue. If you’ve not seen it yet on your T2i, you will, especially for documentary production. We’re not talking about a few jaggies on someone’s tie here. I’m talking about rainbows that’ll make you think you’re tripping on LSD. I saw it on a freakin’ cereal box yesterday.

    More to come…

  5. Darren Says:

    I’m enjoying this conversation, Paul. I bought a Nikon D90 soon after it hit the market and have been flirting with the 7D lately. I don’t have serious filmmaking aspirations, but I’m enjoying the possibilities these cameras offer.

  6. Paul Says:

    @ Darren. I love Nikon lenses, so I really wish they would offer some serious competition to Canon.

  7. Jonathan Poritsky Says:

    @Paul. I don’t miss tape, but I do miss timecode. More than anything, I wish there were a tape, or even a film, analogy that Canon and Nikon (and RED) could sell people on. I like to say it needs to be processed a la film. When people just start to cut an H.264 that looks good on a laptop, things go south fast. I just tried to explain a full workflow for Canon footage to a friend, and when I finished I thought: man, this is so convoluted. (Compressor presets, dig deep to the “don’t screw up gamma setting”, set up a quick cluster, let it cook for a day…)

    @Everyone. Yeah, Canon is whooping Nikon in the video (beat) game, but it’s still annoying that they’re playing it safe in the “SLR” and “Prosumer Video” markets. Where’s our full frame XL-3? With a shutter!? I digress.

    Here’s a better question: where’re the apps? I’m anticipating the iPhone 4 with its 720p capabilities and iMovie and all that good stuff, but the lesson of the iPhone in all electronics is the value of apps. Shortly after the 5D MkII’s introduction, the Magic Lantern project sprung into action to add audio meters and slew of other tweaks to the video interface. Canon can make the gear, but camera UIs have been total crap for a long time, so these folks came in to add the bells and whistles they needed.

    That iPhone analogy again: the apps make the hardware better. It sounds crazy, but maybe I want an NLE in my DSLR. Maybe I want a Holga-crapification widget. Maybe I want to check levels from a bluetooth mixing board recording second system. The point is that we’re already out of the box. Why on earth are in-camera menus roughly the same as they were 10 years ago?

    Off-topic, I know. Hey, don’t I have a blog? Maybe you’ll see some of this land there. One last item: I had been planning to get a point and shoot for my snapping and quick video needs. I’m getting an iPhone instead, and I truly cannot wait for the freedom to shoot and edit in the palm of my hand. And whatever nifty apps companies can come up with for the video aspects, well I’m there.

  8. Stephen van Vuuren Says:


    I’m glad to see your blog back in action as I really enjoy, but I think you missing the forest for the trees in the DSLR issue. My first round of filmmaking was 85-91. I gave it up after losing my shirt as 23 year old with $10,000 (all in film, processing and work print cost) sunk in a bad half-complete 16mm sync-sound film (had to rent the camera, no labs in my area etc.).

    In 1999, I returned to filmmaking spending $10k on complete XL1 kit. I tried all sorts of “film look” stuff but never fell in love with the images it made. In 2004, I spent $8k on a complete DVX100 setup. I really liked the camera and liked the images but wanted more. I sold it when I needed HD for freelance work but did not like any camera on the market, so I bought a HV30 – which I thought made a nice image for the price. DP’d a low budget feature with 3 of them but grew to hate the thing.

    I sold it as soon as the 7D came out (as well as my K10d). My current 7D kit cost me about $2500. I will add a onboard monitor and some other stuff when I have funds – but this equivalent kit to my XL1 or DVX would run about $4500.

    So first, it’s half the price of the DX100 kit – an SD camera with far less dynamic range, color rendition and resolution.

    Second, it’s one of the best still cameras you can buy. The video is basically free.

    Third, while moire exists (the first thing I shot with my 7D was this http://www.vimeo.com/7214842), you are overstating the problem. Plus a little channel blur on the red channel in YUV can often fix a shot. (see my 48 film where I fixed horrible moire on the badge closeup of suspect #3)

    My best 7D work is not yet online – I just was DP on short “12” for Nic Beery (Beery Media) that the rough cut looks great. If you hit me offline, I can send you a link to rough cut.

    You can see my some of my 7D work here http://www.vimeo.com/sv2studios/videos including the 48 Hour film I just shot a couple of weeks ago. 48 Hour films are very tough but audiences respond to these very differently from traditional video. It’s not just bokeh. It’s dynamic range, contrast and color of these fabulous sensors. Canon, Nikon make the best sensors in the world in my book – far superior to most video camera sensors especially under available or less than ideal light.

    The cinematic DOF is just a bonus. The current form factor issues, moire etc. are simply limitations (shooting on film brings limitations). I will close with my email sig that sums up how I feel about the limitations of DSLRs…

    The enemy of art is the absence of limitations.
    Orson Welles

  9. Canon PowerShot SD940IS 12.1MP Digital Camera with 4x Wide Angle Optical Image Stabilized Zoom and 2.7-inch LCD (Silver) | Cmos Digital Camera Says:

    […] DSLRs, Democratic Technology and The Cost of Bokeh: Part 1&#10&#10&#10 […]

  10. Paul Says:

    @ Steven: Thanks, as always, for reading and commenting. I appreciate your defense of DSLRs.

    I think I can see both the forest and the trees, and I want to talk about them both. This post, and the next, are an effort to highlight the risks and trade-offs involved with using these cameras, which I have repeatedly acknowledged are capable of producing stunning imagery.

    The Welles quote is a good one. But, assuming a camera without limitations could ever be invented, I would still be in no danger of having absolute artistic freedom. I work in east Tennessee and southwest Virginia. I have a day job. Even with my funded projects, I work on very small budgets. I work with small crews. These limitations aren’t going away anytime soon.

    For me, because of my other limitations, it makes sense to avoid things that might negate the hard work and problem-solving that I put into my projects. I believe DSLRs, in their current incarnation, risk doing that because of the unpredictable results they produce.

    If it were not for the unpredictability of the moire issue, I likely would not have even written this post. The “jello” effect is one I can live with and work around. The form factor can be worked around. The lack of adequate audio can be worked around. The focus issues can be worked around. That is what this post and the next are about: How to work around the issues.

    But the moire is trickier to work around. The problem is not that it crops up when shooting a test chart, venetian blinds, houndstooth clothing or even (as you did in your test) a shingled roof. It’s that I have seen the issues appear when filming a solid black pique knit shirt, a cereal box, or a dining room table. These are natural things to film and they do not cause problems to other cameras. Furthermore, moire often cannot be detected even when viewing on a small outboard HDMI monitor. You have to view footage on a monitor of significant size (I use a 22″ Apple LCD) after filming to see it. And, perhaps most puzzlingly, sometimes moire appears and sometimes it doesn’t with the same subject. It’s almost as if it was an issue of pure luck.

    At the point that one cannot aim a camera at normal, real world subjects without wondering if it might produce strange effects, well, isn’t that the point that one should question a camera’s usefulness or, at the very least, consider whether what you’re getting in return is worth it?

    That is what this post and the next one is about. Thanks for reading!

  11. Stephen van Vuuren Says:


    I don’t disagree that moire is ugly and potential ruining a shot (though try the channel blur – it really does work to make some shots usuable).

    But I shoot with probably small crew (often just me or me + sound). I have 30+ hours on a doc for a friend in 30 hours, two shots had brief moire and very easy to edit around. We shot 2 hours on our 48 – one shirt had issues but they are still in the film with a little post. Shot 4 hours on the film “12” and not a single shot has moire. Again I don’t disagree it’s bad when it occurs, the incidence in my experience is far lower than you make it out to be.

    Moire is no more a limitation than tape dropouts, rolling shutter, dust in the film gate or the many other issues that exist. And moire occurs on other video cameras as well. Red One and EX1 had issues with black fabrics (IR filter issue). All these problems can usually only be detected by full resolution playback – so thus, there is always a risk when filming with any camera of a shot not being right.

    But with a DSLR, once you shoot a lot, you can figure out if you are at risk for moire and shoot accordingly. Either have a monitor or use DOF/Dutching/etc to insure you get what you need.

  12. Paul Says:

    @ Steve: Drop outs and dust in the gate — good comparisons. I’ve had more moire with the 7D in 5 months than I had of drop outs or dust in the gate in 5 or 6 years of shooting tape-based DV or film.

    BUT — and this is what’s important — it’s working for you (and others) and that’s awesome!

  13. Stephen van Vuuren Says:

    Hmmm. I wonder why you seem to be getting much more moire? It is somewhat lens, exposure settings, overall exposure,shooting style and picture style dependent. It would interesting to see your examples.

  14. Fuel Your Car With HHO Technology? | Technologies Search Says:

    […] DSLRs, Democratic Technology and The Cost of Bokeh: Part 1 […]