Chris Kenny

Anyone considering the 5D Mk. II for motion work should be aware of this. Now, mind you, I've seen some great things shot on that camera, but there are consequences to the tricks Canon currently has to use to get an SLR to shoot at 30 fps. As far as EX3 vs. Red, we own a Red and I've also done a bunch of shooting lately with EX-series cameras. I've done side-by-side tests a couple of times. The Red has substantially less noise when both are correctly exposed, and generally has a much smoother sort of more organic quality to its image -- especially in the highlights. On the other hand, the EX cameras are more light-sensitive and much easier to work with, particularly for run-and-gun style shooting. In short, they're almost completely different tools, and the choice is usually pretty clear cut for any given project.

160 ASA sounds a little conservative to me. I find exposing by eye tends to work pretty well at around 250 for interiors, in Rec. 709 (RedSpace lifts the mids a little more). And I'm very noise-intolerant and not as bothered by some about the way digital highlight clipping works, so I'm sure some people would consider even this too conservative. Of course with the camera recording a linear raw file, there's no real penalty for putting middle gray in the "wrong" place, so there's no reason not to "expose to the right" as much as you can get away with without undesirable highlight clipping (which the camera provides about five different ways to check).

Yes. The sensor has 4520 active horizontal pixels. (There's actually a mode in the latest beta firmware that lets you record 4480x1920.)

Seriously? Still? People have been shooting real jobs with this camera for over a year now, and it seems the same old nonsense is still being repeated by people who've never touched the thing. It works fine. It shoots to slightly modified commodity flash memory. Its power consumption is not particularly unusual by high-end camera standards. The post workflow is more involved than onlining DVCPRO HD in Final Cut. This is not surprising. It's a digital cinema camera, not a prosumer or broadcast camera. I'm not aware of any digital cinema camera that provides more accessible post workflow. You just don't really hear about this with other cameras because they cost so much that few people are attempting do-it-yourself post with them. Even without the contrast patronizing reminders in this forum, I think everyone who considered themselves a serious Red customer was well aware of the fact that the complete system would cost more than $17.5K. The camera system is still vastly cheaper than anything that delivers remotely compatible quality. This is not really even debatable. As for even the most basic post setup costing far more than the camera itself, this is complete and utter nonsense. Red offers far more options for low-cost post than any other camera in its class. You can literally post a Red feature all the way through a 2K DPX deliverable on a laptop. A credible Red workhorse machine is an 8-core Mac Pro with an MXO2, a Dreamcolor, and an eSATA array, total cost under $10K; substantially cheaper than anything that requires you to work with uncompressed HD data end-to-end or ingest any professional HD tape format.

If you count photosites it's a 4K camera. If you look at measured resolution, it's a 3K+ camera. Why would they call it, of all things, a 2K camera?

I haven't seen that episode. But I've noticed in previous episodes that interior car shots -- particularly from the front corners of a car -- tend to look slightly softer than the rest of the show. I think they're sometimes using smaller cameras mounted in the cars for convenience, rather than whatever the rest of the show is shot with.

I tend to doubt you really need 10K to extract all the useful information out of 35mm (at least as shot in the real world), but it's certainly true that producers are having to rethink things. I watched though Firefly (the TV series) on Blu-ray a couple of weeks ago. It was shot on 35, but clearly with SD-only deliverables in mind. At 1080p many shots appear grainy as hell and focus issues are common. The CG shots also don't hold up well at all; they're all upscaled SD and some have other flaws that become apparent on a large screen even outside of that. I'm sure the phrase "Relax, you'll never see that on TV!" was spoken a lot during production. With 1080p being comparable to or higher than the resolution of theatrical release prints, that's not a phrase anyone should be saying anymore.

The 18-85 is delivering now in limited quantities, and is supposedly expected to be out of backorder sometime this summer. And it's one of Red's "Pro" line of lenses; if your judgement of the build quality of Red's lenses was based on the 18-50 or the non-"Pro" 300mm, it doesn't carry over.

I have no particular knowledge of exactly what Red's intentions are, but I'd be surprised if they were planning to turn RedRay into a mass-market consumer device, which among other things would require an interactive layer, elaborate copy protection, and lots of dealmaking with studios. Oh, and a bunch of 4K TVs, to really take advantage of it. I can see 4K computer displays happening in the not-too-distant future, mostly because the extra resolution would be really nice for text rendering. But there isn't all that much of a use case for 4K TVs; most people sit too far away to even really benefit fully from 1080p. Rather, I'd expect instead that RedRay will mostly be a production tool (if things haven't changed, it can also play back Redcode Raw from CF cards and hard drives, making it quite useful on-set, at the production office, in the DP's living room... heh) and an exhibition device of some interest to the festival circuit and art house theaters (being much cheaper and more accessible than DCI-spec hardware, software and workflows).

It's not just that you need to come up with a way to fit information into smaller spaces. That happens like clockwork. The issue is that there are significantly diminishing returns as one moves from highly-compressed to totally uncompressed. The differences between a Blu-ray disc (25-30 Mb/s) and its uncompressed master (989.36 Mb/s) is already subtle enough that most people wouldn't notice them. Once you back off from ~35:1 compression to 12:1 compression (Redcode 28), you've already got footage that holds up pretty well as an acquisition format, as Red has demonstrated. Moving to 5:1 compression will get a little more. Moving to 2:1 compression will get you a very little more. And moving from 2:1 to uncompressed gets you nothing or almost nothing. At the same time, at a given data rate, a compressed signal is (assuming you're not misusing your compression algorithm) always going to look better than an uncompressed signal that has been downscaled or (especially) acquired at a lower resolution. Conceptually, in this context, the best approach is to think of downscaling as an extremely primitive and ineffective compression algorithm, which is being compared to much more advanced algorithms. The same is true of saving space by reducing bit depth, subsampling color, etc. Combine these two factors and you can derive the criteria under which uncompressed makes sense. It doesn't really make sense until you have enough resolution, etc. that nobody could reasonably want more (or no sensor can acquire more) and technology has advanced to the point that it's almost trivial to work with such footage uncompressed. For instance, uncompressed SD is nearly trivial now. But compression hasn't died yet, because nobody was happy with the resolution provided by SD; they want HD (or better), and uncompressed HD is still annoying. At some point, people will presumably decide that 16-bit-per-channel 8K (or whatever) is really enough. And at some later point, advances in storage technology will make it practical to work with that format uncompressed. And only then will lossy video compression start to disappear.

I can't imagine any technical reason why this codec couldn't be used for material from other sources. There was some info about the post process for the reel over at RedUser; the compressed files were produced from DPX and TIFF files. Once Red footage has been graded and rendered out to DPX or TIFF or whatever, it's just like any other footage in such files. Though remember that noise and grain are the enemy of aggressive compression. Correctly exposed Red footage is pretty clean, and I suspect the result benefited from this. It's not too hard to imagine you might have to up the data rate a bit if your footage has a noise/grain texture that you're interested in preserving.

The 1080p trailers on Apple's trailer site hold up pretty well on my 46" 1080p LCD from a viewing distance of about 70", which is close enough that practically everyone who sees my living room asks why I have the couch so close to the TV. They also hold up pretty well on the 92" screen in the screening room at work (with a 1080p projector pointed at it). Are there more noticeable artifacts than with a Blu-ray disc with 2-3 times the bit rate? Yes. But they're certainly quite watchable, and I rather suspect I'm more picky than most people. 10 Mb/s is low enough that I wouldn't be too surprised to see it creep up slightly before the product is finalized (Redcode Raw's data rate crept up a bit from the initial announcement). And I'm sure there will be detectable artifacts under close scrutiny if you know what to look for (which most people notably don't, with wavelet codecs). But nothing Red is saying is so far out there, technically, that its honesty should be called into question.

Karl, the RedRay is a player, and this 10 Mb/s 4K codec being discussed is for delivery/playback, not acquisition. Solid state is pretty clearly the future of acquisition, at Red and elsewhere. (And I'm curious why you say it's unreliable. That certainly hasn't been my experience, and SSDs are being regularly deployed these days in data centers in extremely mission-critical applications, so the IT industry clearly considers solid state storage quite reliable.)

Not sure why some people are so skeptical of this. You can get pretty good looking 1080p at 10-11 Mb/s with H.264 (see trailers on Apple's web site), while MPEG-2 requires 6-7 Mb just for relatively clean 480p. This means H.264 is in the neighborhood of 4x as efficient as MPEG-2. So if this new Redcode-derived delivery format does decent 4K at around 10 Mb, it would just be to H.264 what H.264 is to MPEG-2. Does this really seem that impossible? Or even, framed in this context, that unexpected? Of course, there's no such thing as a free lunch. I'm sure it requires gobs more processing power than H.264, just as H.264 requires gobs more processing power than MPEG-2. I suspect real-time software-based decoding is a long way off. Fortunately, this is the sort of problem that can be solved with a relatively inexpensive ASIC, these days.

You appear to have missed my point. My point is that HD distribution raises the bar for quality. Your image quality is now a liability if your footage won't hold up at 1080p, whereas previously for non-theatrical distribution, your footage only needed to hold up at 480p. And the move to HD generally in the consumer viewing market has put more emphasis on image quality -- it's now more of a selling point than it used to be. So, the fact that distributors previously bought all sorts of stuff shot on lower-end formats doesn't demonstrate that image quality isn't a significant factor in the current market. As it happens, no. That was Tenolian, with his comment that "A movie shot on Pixelvision starring Brad Pitt has an infinitely better chance at distribution than a movie shot on 65mm without any known star actors." My point here is that while this may be true (well, if one looks past the hyperbole), it's completely irrelevant to the subject of whether shooting on a higher end format makes your movie more likely to get distributed. It's only relevant to a situation where you have to choose between shooting a film with no-name actors on a Red and shooting a film with Brad Pitt on a DVX, and I'm not aware of anyone having ever been in a situation where they had to make that choice.

They're irrelevant to most low-budget filmmakers because regardless of how useful it would be for them to attach a known star, they have no realistic chance of doing so.

If you're sticking to strictly theatrical releases, I wouldn't characterize "many many many" as accurate. And the standards for getting non-theatrical distribution are changing as a result of consumer HD formats, so pervious success in that arena doesn't necessarily mean anything, looking ahead. Anyway, the hypothetical is useless. The fact that a movie featuring Brad Pitt will sell almost regardless of any other factor is completely irrelevant to most low-budget filmmakers, who have no realistic possibility of attaching A-list talent to their projects. In other words, whether or not shooting format has a substantial impact on the commercial possibilities for the latest Brad Pitt vehicle says absolutely nothing about whether or not it has a substantial impact on the commercial possibilities of a low-budget film featuring unknown or lesser-known talent.

Sure. There are some movies which, conceptually, work just fine (or even benefit from) being shot on a format that doesn't quite meet the quality standards of most commercially released features. But major distributors are never going to have an appetite for more than a couple of them a year, and a lot of filmmakers are going to be interested in pursuing concepts which don't happen to work well that way. I think (at least in the developed world) we're going SD DVD fade away in favor of some mix of Blu-ray and online distribution in about the same time frame it took DVD to overtake VHS, meaning DVD will be overtaken perhaps around 2013. Exactly what mix of Blu-ray vs. online we see is going to depend on a bunch of factors, and will probably vary from region to region based on the local quality of consumer Internet infrastructure. But either way, my point is basically that, whether via Blu-ray or online distribution, people are going to be watching most content that was acquired in HD (or on film) in HD in four or five years. Distributors mostly know this. If you're shooting in a format that doesn't hold up all that well at 1080p, that is going to have a real impact on the commercial success of your project. Pixelvision might be a bit of an exaggeration, but I take your point. Thing is, any talent big enough to substantially increase the odds of commercial success for a film is going to cost far, far more than the price difference between (say) shooting on an HVX and shooting on a Red One. So at least until you reach the level of the Red One (virtually everything below it is badly compromised in terms of compression format and/or resolution), spending more on camera does seem to be a worthwhile investment.

A few points: 1. Rental prices on digital gear aren't 1%/day. They're typically in the 3-5% range (maybe a little lower on very expensive digital gear), reflecting the fact that it gets replaced more often. Red rental rates have settled at about this level, and there's no particular reason to think they'll go much lower. 2. Those declaring that Blu-Ray is a dead-end are judging by a standard that has no relationship to reality. It took DVD six or seven years to pass VHS, and that's considered one of the most successful consumer format adoptions in history. It's way too early to say Blu-Ray is a failure. And if it ultimately is, it won't be because people will stay with SD DVD forever. It'll be because people transition to digital downloads, many of which will be HD. 3. Obviously anyone who thinks their movie is going to be a success because just they shot it with a certain camera isn't thinking too clearly. But this is not the same thing as saying that acquisition format doesn't matter. If you're trying to shoot a movie that might actually get distributed, and it doesn't hold up on the big screen or at least at 1080p (which, remember, is at least as much resolution as a typical theatrical release print), you're not even in the game (unless there are some uncommon extenuating circumstances).

Probably not quite. You're probably looking at effectively 1.6K luma resolution in 2K mode, and chroma resolution maybe in the 1K range or a little higher. High-end 3-chip 1080p broadcast cameras are probably delivering something more like 1.8K for both luma and chroma. (Of course, if you're recording to HDCAM rather than HDCAM SR, you're only recording 1440 horizontal pixels, which would even things up a bit.) As others have said, Red's 2K windowed mode is nothing to write home about. There's a reason they're really pushing for higher frame rates at full resolution with their next generation of cameras.

Assuming the FF35 Epic has 6144 horizontal photo sites (6*1024) and is 36mm wide (standard full frame), you'd have 4247 photosites across a 24.89mm frame. I would guess for the sake of simplicity the windowed mode would probably just use 4K (4096) horizontal photosites like the Red One.

Presumably because it gives you more options. With the larger sensor, if it's a shot where pulling focus isn't that challenging, or where there's plenty of light (so you can stop down without underexposing), you expose normally and get a better image than you could with the smaller sensor. If it's a shot where you haven't got enough light and there's going to be a tricky focus pull, you stop down and put up with a more noisy image -- but an image that probably holds up fine relative to what a smaller format camera would deliver when exposed normally. So you get a camera that produces better images under ideal conditions, and similar images under more challenging conditions. If the numbers from the original Nov. 13th announcement are still accurate, Monstro (the FF35 sensor) will deliver about two more stops of dynamic range than Mysterium-X (the S35 sensor). This presumably comes down to its larger photosites. Given the constraint that you need at least 5K+ photosites to deliver 4K+ debayered resolution (which is clearly one of Red's goals), you obviously have no choice but to make the sensor physically larger if you want to make the photosites larger.

Remember, the camera shoots to a raw format. The recorded image contains data all the way down into the sensor's noise floor. With most other cameras, your blacks are to some extent already crushed by the camera's on-board image processing system, so you don't see how much noise was actually there. With the Red One, you can choose to crush the blacks and kill some of the noise, or keep the shadow detail and live with the noise. The Knowing trailer contains a bunch of night shots, presumably shot on the Red One like the rest of the movie, though of course it's impossible to know exactly how well lit they were. (They don't particularly look like available light.) We have some decent-looking low-light footage (mostly from test shoots) shot with Red 404 + Superspeeds. If I remember I'll grab a few frames when I'm at the office this afternoon.

Well, remember, a larger sensor means either larger photosites (lower noise), more resolution (also lower noise, once you scale to your deliverable resolution), or some combination of the two. This means you should be able to underexpose (stop down to get deeper focus) and still retrieve a decent image. Or maybe this will be the impetus for someone to finally build a practical and affordable autofocus system that works for cine-style shooting. Some kind of targetable rangefinder system married to a focus control motor and a lens data system, presumably.

Well, they appear to have set platforms other than OS X as a low priority, anyway. Which, I have to say, I honestly can't bring myself to care about all that much.