Tuesday, March 30, 2010

The DSLR Revolution

If you happen to be like me, a total camera nerd (or at least an enthusiast), you’ve probably been following some of what’s been going on with new DSLR still cameras that are capable of shooting HD video. These cameras are now being referred to as HDSLRs. Some of you, however, may not be following this quite so closely, or may have heard of the concept but are wondering ... what’s so revolutionizing about DSLR cameras that can shoot video?

Allow me to briefly explain. I want to start with the most obvious--and yet most totally overlooked--aspect of a good camera. The lens. The glass you shoot your image through is really at least 70% of how good your image is. Forget about all the stuff behind the lens, and those great digital sensors that are amazingly light sensitive and have great resolution. All of that is naught if you’re lens is junk. Want proof of concept that lens really is so important? Go look up the price for the Red One camera. Then look up the price for a PL-mount professional grade cinema lens by, say, Carl Zeiss. A great prime in that category will easily cost as much as the $20,000 to $30,000 Red One set up (depending on how you customize it). Try buying a set of such prime lenses, and now you’re talking about the price of a house.

Now, we’re not dealing with those kinds of lenses. I just wanted to prove a point here. But the first revolutionary idea these new HDSLRs present to us is the option of shooting with some great still photography lenses. Depending on your budget, you can easily spend between $100 and $5,000 on a single lens for a DSLR. What’s more, renting still photography lenses is really quite affordable. I’ve done it for two projects (back when we were shooting with lens adaptors that allowed use of still photography lenses on a regular HD camcorder).

In fact, quite some time ago I wondered why no one was making a video camera like the HVX200 that could shoot HD video but simply had a mount for still photo lenses. What I didn’t realize at the time was the issue of sensor size. DSLR lenses just aren’t designed to work with a 1/3-inch chip that most prosumer cameras have. It’s just far too small!

And then lens adaptors came along. And I have worked closely with those, though I never bought one. And here’s what I have to say about lens adaptors. At the time, they were a gift from God. You could use still photo lenses and get great depth of field, achieving a wonderfully cinematic look for your project. However, they also were a total pain in the ass. They suck up light. The last project I DPed with a lens adaptor, I tested and rated the set up to be the equivalent of 50 ISO film speed. For those of you who understand ISO (or ASA) numbers ... you know that that’s about as slow as film stock comes. The lower the number, the more light you need. Most people shoot still photos around 400 or 800 ISO. My new DSLR doesn’t even drop bellow 100 ISO. So, when shooting with the adaptors, it always felt like we had to set our sets and actors on fire to get an image at minimum exposure. We also had to shoot with the DSLR prime lenses wide open, which didn’t allow for them to be at their peak performance (which is usually about one or two stops up).

The other problem with lens adaptors is that they were noisy--and I mean the image. The system works by allowing the DSLR lens to focus on a ground glass that has grain. That ground glass vibrates or spins. The video camera is focused on the ground glass. Thus, the ground glass stood in for the larger chip size needed to use DSLR lenses. But you definitely had grain. Now, it managed to look pretty organic, like film grain. But even shooting with a 1080p camera, I think you’d be lucky to have an image shot with a lens adaptor that looked no better than 16mm film once it was projected on a movie theatre screen.

But here we are, and DSLRs are shooting HD video. In fact, it’s been going on for more than a year now. Slumdog Millionaire effectively used the Canon 5D Mark II to shoot sequences of the film. Since then, other models have come, and there’s been a lot of good competition, though frankly Canon has lead the way, with Nikon lagging behind only managing to put out 720p capable cameras (aside form one that shoots 1080p but only 20 frames per second--what the heck is that?). In fact, Red was forced to re-evaluate their Scarlet camera and they went back to the drawing board to make a Scarlet that will compete not just with the prosumer video camera market, but the DSLR market as well. So far, however, I have this sinking feeling Red’s letting things get away from them by continuing to push back the release of the Scarlet. Meanwhile, more HDSLRs hit the market at lower and lower prices, making the prospect of buying a Scarlet for guys like me more of a ... “meh, we’ll wait and see what it does compared to what I have” kind of thing. Of course, it will be worth keeping in mind that having worked with the Red One I believe the Scarlet still stands a chance to really blow us away.

Now, I don’t want to get too drawn in to discussing the specifics of these HDSLR cameras. If you are interested, here’s a great rundown of features and what to look for in a camera: http://nofilmschool.com/dslr

But I do want to touch on the other features that make these cameras so fantastic. The CMOS chip inside the Canon 7D and the T2i (the latter of which I own), is in fact just a smidge bigger than the sensor inside the Red One. And if you spring for the Canon 5D Mark II, the chip is a full 35mm frame size, a good bit bigger than the Red One’s M sensor.

Now, I’ll speak from my experiences here. I own the Rebel T2i (aka EOS 550D). It just hit the market. In fact, Canon moved up the shipping date by a week, so I got it sooner than expected. In fact, just in time for it to be 2nd camera on a short film I DPed with the 7D. The guts of the 7D and the T2i are pretty much the same, and having spent last weekend shooting with both cameras side by side, I cannot honestly see a difference in picture quality. We shot in 1080p24, and I’m impressed with how clean the image looks and how effective these cameras are in lower light. Now, compare the 18 megapixel CMOS sensor of the T2i to my Sony FX1 prosumer camcorder with something like 2 megapixels, and you start to see that my new T2i, which cost a fraction of what my FX1 did when I first got it, is miles ahead in capturing a beautiful image.

And that's the final revolutionizing aspect of the DSLR cameras is their price. Compared to the prosumer HD cameras out there, most of these HDSLRs are cheap. So, for the filmmaker on a very tight budget, buying the T2i for $800 and getting a couple lenses for another two or $300 sounds a lot better than spending six, seven, or more grand on a video camera with a fixed zoom lens and a small sensor that makes getting shallow depth of field next to impossible.

Having said all of this, there are drawbacks to the DSLRS. They are tools like any other, and one needs to know their limitations. For the shooters wanting to cover long live events, ENG style shooting on the go, and hand-held documentary work, HDSLRs present more challenges than advantages. For that kind of work having a shallow DoF is actually kind of a pain, and potentially a hindrance. With my T2i, I cannot zoom while shooting video without having noticeable changes (as in jumps) in exposure due to the lens’s iris adjusts as I zoom (a personal note here: when it comes to narrative filmmaking ... I hate, hate, hate zooms anyway!). Takes are limited to a max of 12 minutes because the maximum file size the camera can create is 4 GB (or 12 minutes of 1080p HD). The cameras are also small and pickup micro-jitters when working hand-held. So you need some shoulder mount system or counter balance to smooth things out. And the video is quite compressed using the H.264 codec, which as of yet isn’t really supported natively by most editing systems. I transcode to Apple ProRes (regular or HQ depending on the project) before editing.

Now, all those drawbacks in mind ... if you are an indie filmmaker trying to make your next short film or low budget feature, I think you really need to take a look at the options and advantages an HDSLR can bring to your next project. The cinematic look one can achieve is spectacular. 

But allow me a final caution. Again, these cameras are just tools like any other. They won’t do the work for you, or magically make you a great cinematographer or filmmaker. Yes, they will shoot a very nice image. But you still must know the cinematic language of shot composition and selection, good lighting, editing, and aesthetic style and storytelling craft. 

I think there’s a temptation among many new filmmakers to look at tools like this and think, “That’s it, I can now make my movie and it will look amazing! People will love it! I’ll get into Sundance! I’m on my way!!!” This happens every time a new breakthrough in cameras happens. When HD first came into the prosumer market, it happened. I witnessed it personally and was even party to it. The truth is, these are just tools. Tools can be used well and they can be used poorly. These are also not perfect tools. They have definite limitations. 

So I caution new filmmakers to keep these things in mind: A camera that can shoot beautiful footage is no excuse for not properly lighting a scene. A camera that can shoot beautiful footage doesn’t make up for not knowing how to shoot for your edit. And ultimately, nothing replaces or makes up for bad storytelling. Wanna see a great film shot on a less than fantastic camera ... watch Once! Always make your tools servants to the story you are telling, not the other way around.


Jack the Mack said...

Thanks for the way cool and informative post on the DSLR craze. Really fancy obtaining one.
One thing I don't get: why do DSLR's not need the vibrating / spinning ground glass? I understand why hacked / modded video cameras (i.e. cameras that have a lens adaptor attached) need the moving ground glass. Why not DSLR's? They have ground glass, right?

Mikel J. Wisler said...

Hi Jack the Mack,

Actually, no, DSLRs do not have a ground glass. They don't need a ground glass. The lens shines its light directly onto the the CMOS digital chip inside the camera, the same way a conventional video camera captured light from its lens and shines it on it's smaller chip to converts it to data. This is exactly how film work too: light enters the lens which focuses the light on to a frame of film. But with a DSLR, the CMOS chip stands in for the film and the process if digital instead of photo-chemical.

The difference between most video cameras and video enabled DSLRs is that the chip inside a DSLR is much larger, allowing for a much more cinematic look (i.e. shallow depth of field.) So the beauty of using a DSLR for video instead of an adapter with a ground glass is that the ground glass both sucks up light and adds grain to the image where as the DSLR can skip this step and still get great looking cinematic footage.

Thanks for the question. Hope this helps!

Jack the Mack said...

Thanks Mikel

Weird - I always thought movie cameras had ground glass in them: the more you think about, the more it makes sense NOT to have it in a movie camera. Ground glass (GG) is used in an SLR to VIEW the image, right? It's like a screen that the image (reflected from a mirror) is projected onto, and that image is quite bright, like a cinema screen. The image sits on the ground glass. A
But why does wikipedia say the following about movie cameras: (from http://en.wikipedia.org/wiki/Ground_glass )
"In motion picture cameras, the ground glass is a small, usually removable piece of transparent glass that sits between the rotary disc shutter and the viewfinder. The ground glass usually contains precise markings to show the camera operator the boundaries of the frame or the center reticle, or any other important information. Because the ground glass is positioned between the mirror shutter and the viewfinder, it does not interfere with the image reaching the film and is therefore not recorded over the final image, but rather serves as a reference for the camera operator.
Ground glasses commonly serve as a framing reference for a desired aspect ratio. Because most films shot with spherical lenses are shot full-frame and later masked during projection to a more widescreen aspect ratio, it is important not only for the operator to be able to see the boundaries of that aspect ratio, but also for the ground glass to be properly aligned in the camera so that the markings are an exact representation of the boundaries of the image recorded on film."
If I follow this, in a movie camera it is used for framing. It's removable. Is it taken out for each shot?
Sorry for all the questions.
If I have this right, putting GG in an adaptor and then fitting it onto a video camera emulates a bigger chip, because it creates a bigger viewing area...? The image comes through the lens, and "sits" on the GG. Why is this desirable? Why don't video camera makers put a bloody piece of ground glass into their camera? Maybe they're planning it...
; )

Thanks again Mikel...

Jack the Mack said...

Oh, another thing: if I'm reading it right, it's the SENSOR that seems to be the most important thing.
So hypothetically, in a parallel universe, if those video cameras (that people have been putting adaptors / spinning GG on) had always had a really nice big chip in them, those adaptors would not really be needed? They would get a cool shallow depth of field just because of this imaginary huge sensor? (I saw a picture the other day of the unbelievably HUGE chips RED are planning on putting in the Scarlet)...
Is a sensor and a chip the same thing?
Another thing: in 3CCD (3 chip) cameras, how does the light from the lens fall on THREE chips?
You can invoice me for the tutorial. I owe ya. ; )

Mikel J. Wisler said...

Hi Jack the Mack,

Yes, you got it right that there is a ground glass element for the viewfinder in cameras, but that is a different type of ground glass than what is used in the adapters since it is used for references in framing the shot in the view finder (and is just a permanent part of the camera that doesn't have to be removed during shots). So yeah, that type of ground glass is definitely there. Good job!

And you have the right idea about the ground glass in adapters. What they do is effectively make the "frame size" that a conventional video camera is attempting to shoot the size of a 35mm film frame, thus why the depth of field can be so good. The way such adapters work is that they allow the camera to focus on the image created on the ground glass because their 1/3" chips (aka sensors, you are correct, those are the same thing) can't quite generate that kind of shallow depth of field.

Now ... I wouldn't say the chip or sensor is the most important thing. Honestly, the lens is still more important. There's more to image quality than just having a shallow depth of field, and how good the lens is determined most of that. The two big things a sensor helps determine is resolution and depth of field. But both of those are directly contingent on how good the lens is. You can have an awesome full-frame 21 megapixel sensor like the Canon 5D Mark II, but if you slap a really bad lens on it, your images will still look bad (potentially worse as the 21 mp sensor might amplify the crappy-ness of the lens). So the combination of good chip good lens is important for professional grade cinematography.

Now, what you described in this hypothetical parallel universe--FRINGE reference :) --is in fact happening. Panasonic and Sony are making larger sensor video cameras. However, there are reasons they haven't so far ...

1. Auto focus on a camera with such shallow depth of field is really difficult. Much easier to accomplish on a small 1/3" or 1/2" chip and zoom lens. And for shooting live events like sports or on-the-go documentary work, shallow depth of field can be a hindrance instead of a help as you spend too much time searching for your focus manually.

2. These large chips get hot! So, for small video cameras, it's really hard to keep them cool enough to keep shooting for a long time.

3. Lenses made for most prosumer and professional video cameras aren't made to fill large sensors.

So, there's pros and cons. For professional grade narrative filmmaking, these DSLRs present options for those on a budget or looking for new ways to shoot projects that benefit from small cameras (like the final episode of HOUSE this season that was all shot on the Canon 5D Mark II with Canon lenses).

But you are right that now that HDSLRs are on the market and that video cameras with larger sensors will soon hit the market, yes ... lens adapters are in fact obsolete.

Great discussion. Thanks for reading! Let me know if there's any other ways I can help. Hope this has in fact been helpful.

Jack the Mack said...

Thanks for that Mikel..
I figure that soon those crews that make all those expensive adaptor kits will have to, er, adapt.
I've picked up more information here in a few minutes of concentration than tens of other labyrinthine blogs I've looked at... so thanks for replying, and for your lucidity...

Mikel J. Wisler said...

Thank you very much, Jack! I'm glad I was able to help. Some times these technical discussions can run in circles and be a bit hard to follow, but I'm glad you were able to get something from my blog and responses. I certainly enjoy interacting with readers!

Have you checked out my entry on actually using the Canon 7D and T2i for some film projects I shot? http://mikelwisler.blogspot.com/2010/05/7d-and-t2i-cinematography.html

I've also been posting things to: http://dslrvideo.6te.net

Again, thanks for reading and don't hesitate to let me know if I can be of any more help!