RED Addresses the "DSLR Market"

“If people are not stunned by the specs and design, I will retire… truly.”

-Jim Jannard, RED

Jim Jannard of RED had cryptically mentioned that they would be creating a product to address the “DSLR Market.” Last night on REDuser.com he expanded on that announcement a bit:

We believe, and are developing for late 2009, a replacement for DSLRs. Currently, we call it a DSMC (Digital Still & Motion Camera).
While (insert code name) is not a replacement for Epic or Scarlet, it is strategically targeted at the DSLR space. As Nikon and Canon release their 720P and 1080P, respectively, DSLRs with video capture… RED has a more advanced view of the future. We look forward to rapidly pushing the “big guys” along in feature sets and capabilities.
RED firmly believes in higher resolution, higher S/N, higher DNR, higher frame rates, smaller bodies, more system flexibility, and many more options as we move forward in camera development.
The strength of RED is in our sensor development program, REDCODE, and having no legacy platforms to deal with. That left us free to explore, develop and prepare to deliver a new platform. DSMC.
We think all our customers already know what the future will bring. They are just afraid to wish for it for fear of disappointment. Fear not. Sleep tight. RED is awake.
Jim

Jim’s enthusiasm-trumps-accuracy grammatical stylings are oddly representative of his company’s engineering efforts. As Fred Johnson pointed out on TWiP 46, RED is a lot like Apple entering the cell phone market. It takes an outsider to start over from scratch and shake up the assumptions that the staid players hold dear. But as an outsider it’s very difficult to just get the basics right. RED One was both groundbreaking and revolutionary, but in many ways it is still catching up with the things the “big guys” do in their sleep. Just like the iPhone.

Jim didn’t say that RED is building a DSLR. He said he’s making a replacement for DSLRs. I love my DSLR but I do have the sense that much of its design is based in a legacy that no longer applies. Jim also said on CML that “We also believe that the future is a ‘still & motion’ camera world. Just at much higher quality levels [than the D90].”

What makes up the user experience of a DSLR? To me it’s more about lens choice, a big sensor, instantaneous shutter release and a raw workflow than it is about mirrors and Through The Lens viewfinders. It’s about insanely good autofocus and lenses around which one feels compelled to form a religion. It’s about ergonomics that make you forget where your hand stops and your camera starts.

Add motion to that mix and you get to list off more personal priorities. The Nikon D90 has been amply tested by many folks and the verdict is just as I predicted: The images look great as long as you don’t look to close, and motion is fun but too much turns into jello-vision. Nevertheless, the D90 got us all thinking about what it would mean to dip our SLR chocolate into our HD peanut butter, and for my part at least I couldn’t think of a good reason not to do it. If RED or anyone else can develop a raw motion workflow in a body that not only shares my SLR lenses but can do the job of my SLR, then Jim is not wrong in saying that “‘revolution’ applies more to this than the RED ONE did to cinema.” I mean, except for the grammar.

But while the “big guys” in the SLR market may be set in their ways and unlikely to revolutionize without competition from someone like Jim, I hope Jim realizes what he’s up against. I think he should, because he’s an avid photographer. He’s going down a whole new path now, competing with well-loved players instead of the rarefied few in the digital cinema world. Canon and Nikon compete viciously on features, price, and technology. It’s a heavyweight bout where the spectator is the only clear winner.

There is a theory that in order to shake customers away from an existing product, your product must be ten times better. In the digital cinema space RED One was seen by almost everyone as being ten times better than anything at its price point. What happens when RED’s SLR-killing autofocus is just a hair slower than Canon’s? Or only a tiny bit better?

Good luck Jim. I’m sure I’ll want whatever you make. But I do hope you’re as busy making things as you are dreaming them up. Remember the 4K projectors and displays you mentioned two NABs ago? It would delight me to no end if you were as enthusiastic about refining your existing products as you are about announcing new ones.

DSLR Movies, Pros and Cons

In my new tradition of rambling on about a subject only to post again a few days later with a more succinct summary of my thoughts, here’s a quick rundown of why you should be excited about shooting video with your DSLR, and why you should reserve some modicum of wait-and-see caution.

First, the reasons to get excited, ranging from the obvious to the more obvious: 

  • In the case of the D90, what a sweet deal. $1,000 for the body. I paid nearly that for my HV20. The real win here is that a resolution that is low to medium for a DSLR is positively overkill for HD. Nikon’s “bargain” SLR is overqualified by a mile for 720p video.
  • Use your DSLR lenses. If the 5D MkII (or whatever) shoots video too, then both Nikon and Canon peeps can rejoice about this one. A big drawback of the RED One is the expense of the lenses (which need to be top-notch), and a perceived drawback of next year’s Scarlet is the fixed lens.
  • A big-ass sensor. The D90’s sensor is roughly the size of the RED One’s. The rumored 5D MarkII’s is way bigger. That means more control of depth-of-field and more predictable results from your stable of lenses. It also means that a $1,000 camera is now making images that, at web resolutions, look an awful lot like those from much more expensive kit.
  • 24p. I’m still reeling from this one, but somehow the D90’s video wound up being 24 fps. Hallelujah. It so easily could have been anything else.

And now the reasons to reserve judgement:

  • 720p only, at least in the case of the D90. Personally I love 720p, but it’s quite a subset of what your DSLR can do.
  • 24p. And that’s all, on the D90 anyway. No overcranking or undercranking. And who knows what frame rates other DSLRs will offer. A camera on which video is an afterthought is not likely to offer a wealth of options here.
  • CMOS. CMOS roll. Roll MOS roll. Are DSLR chips, which have never had to fear shearing, skewing and wobbling from rolling shutters (mechanical shutters negate this), going to fare well when recording video? Or will they be jello-cams?
  • Limited running time. Aparently there’s some red-tape reason why the D90 is restructed to five minutes of video. Those five minutes will still cost you about 600MB.
  • No external audio input. But the D90 does have a mic. You’ll be dual-systeming it. Fortunately there’s cool software out there for syncing audio based on waveform matching.
  • Manual control. Although things look good for the D90, the thing about shooting video is that you need the same kind of quick access to key manual exposure controls that SLR stills shooters require. But will the video options be ghettoized in a deep menu?
  • Manual focus. Not in and of itself a problem, since that’s how pro video and film gets shot, but that LCD screen won’t be reliable for critical focus at HD res. HD cameras have handy focus-assist features like edge enhancement and LCD zoom.
  • You’ll be at the mercy of a codec. The D90 scores well here on paper, but you are still dealing with heavy (and inefficient) compression of a baked-in color palette that may not respond well to agressive grading, as tends to be the case with perceptual compression. This is a tough thing to swallow when you’re holding in your hands a camera that lives and breaths raw formats for stills. Maybe someday someone will make a DSLR that lays down CinemaDNG sequences. Maybe? How about Please God Yes.
  • But maybe worse than baked-in color and codecs, you’re at the mercy of whatever the camera can muster in realtime debayering. Most video from compact cameras looks bad not because of compression but because of the hasty techniques used to rapidly build an image from a tiny subset of the sensor’s photosites. I offer up my beloved LX2 as an example—its videos may be 848x480 on disk, but what’s actually present in the images is far less resolution than that. This hard-to-quantify factor could be the real pitfall, although none of the (heavily recompresed) D90 sample vids I’ve seen have exhibited egregious demosaicing artifacts.

I said I’d stop hypothesizing, but this is an interesting enough subject that I felt it was worth clearing the air about both why this thrilling new trend in DSLRs is so great, and why it’s not as great as one might hope.

UPDATE: Mike commented below with a link to dpreview’s sample D-Movies. If you download the original AVI files you can see some sizzle and color sparkle that is symptomatic of the expedient debayering I describe above.