HD For Indies

OK folks, let's do this: about 10,000 words on SI-2K vs Red One, no holds barred.

...so I've just written two articles after being offline under the weather for several days, and it just cries out to do a little pros & cons, A vs B kind of a thing here. So let's do it:

UPDATE: scroll to bottom for latest updates, there's some significant new info

Silicon Imaging's SI-2K camera vs. the Red One camera

I've talked extensively with both teams, but if you're a regular reader you're obviously aware that I've spent significantly more time with the Red team, even working two tradeshows with them. If anybody has a valid counterpoint on anything, comment away and I'll incorporate the good ones. Now, keep in mind, that both of these systems are not shipping yet, and I'm basing my commentary and analysis on what I've seen and been told by both developers. So this is a preliminary analysis based on statements to date, the situation and analysis may change over time.

I'm going to take a look at a variety of issues:

Project Status
Form Factor
Resolution Options
Frame Rates
Sensor Size
Recording Modes
Audio
Outputs
Lens options
Codecs
On Camera User Interface
Image Quality
Workflow in Field
Workflow in Post
Price
Financial Backing & Partnerships
In Closing

This is my sit-down-and-type-nonstop draft, let's start with the basics:

Project Status

SI-2K: they say they're going to ship the MINI head in December, and the MINI + DVR (essentially the whole camera) in January. They've been showing a working prototype since NAB, and the first feature is already in post, and they seem pretty confident and firm about ship dates.

Red One: we've seen footage, we've seen non-functional prototypes, they say they'll have working prototypes by the end of this year, but don't expect to start shipping for real to paying customers until something like the March/April timeframe...with the usual "all subject to change, no promises" caveat. Although Jim Jannard, founder of Red, is not one to claim a date if he isn't pretty confident about it. So their intent is solid, we'll have to wait and see if they are capable of shipping in that time frame.

WINNER: SI-2K - they're clearly further along in their development arc of their camera as compared to Red - unless something significant happens for either team (which I doubt at this point), they'll ship first by months. Long term, this isn't hugely significant, short term, it matters if the camera is available for your project or not.

Form Factor

SI-2K: final form factor has yet to be seen - they're still working on it, and I'm expecting something different to pop up from what we've seen.

But in terms of approach, they've got the equivalent of the Star Trek Next Generation's saucer section - they can break their camera into two parts and send each one where they want it (yeah, first of many geeky analogies). The SI-2K has two parts - the MINI head is just a 65mm wide, palm sized block with a B4 mount on the front, power and GigE connectors on the back. It is TINY. Tiny can be a beautiful thing - it can essentially just be a small appendage on a lens. I have pocket sized converter boxes bigger than that. Just like in Star Trek, the saucer section (MINI head) can run off into battle, leaving the bulkier, heavier main section behind to not get shot at. Except that here you'd need a gigabit ethernet cable running back to the computer, and battery power for the MINI head. But GigE is lighter, more flexible, and cheaper than HD-SDI grade coaxial cable, so that's a total win. It is the uncompressed RAW signal running through that cable back to the computer, which can either compress it with Cineform RAW or just record it uncompressed. Schweet.

Meanwhile, the DVR (digital video recorder) can either be attached or up to 100 meters away (or further with a fiber converter and rig) - so somebody can be sitting comfortably at a desk and monitoring it from there, controlling recorder functions, etc. But if you want to do that, you'll want the touchscreen up by the camera head, so a second ethernet cable can be run (with a VGA to Cat5 adaptor, such as those made by Gefen) to bring the touchscreen back up to the operator. Two ethernet cables aren't much to wrangle and can take a good deal of abuse, and are readily available as well.

In "normal" mode I expect it will be a somewhat large camera since it has to incorporate an Intel motherboard (I'm guessing laptop based?) and a graphics card - we have to wait and see what they come out with for their next and presumably final rev.

So...I don't know how small or what shape the DVR is going to be, but we know it has to include an Intel motherboard and some cards, so it can't be too small. That's the trade-off with the camera - to get all the benefits of running a straight-up WinXP app with the slick UI and features, it is stuck with that beefier form factor. Dunno on the weight either, gotta wait and see.

So - for remote head operations it'll be great, but for steadicam or run 'n gun, not so great I'd guess, but we won't know until we see the final form factor.

From a market positioning standpoint, splitting the product into two clear, distinct parts was genius to differentiate themselves from Red in a way Red can't respond to in kind.

UPDATE Ari Pressler of Silicon Imaging wrote in to point out an oversight on my part:

I think one very powerful configuration is missing in your analysis: The  Core 2 Notebook + MINI.
 
Core 2 Notebook ($2K) + MINI ($12.5K + Accessories)  = 2K Recording,  look editing station and HD monitoring in Component and DVI for direct projection.  If you have a fast enough disk system or enough RAM buffer you can even record 12-bit content, export to DNG (or DPX sequences) and use a fully uncompressed workflow that you already own (no additional cost)!  Plug in USB drives for additional mags for the local computer retailer on the corner.
 
I can shoot on the laptop, edit and transmit/network all from the same device!!
 
Wireless access to the recorder allows remote desktop operation or exchange of .look file information which can be generated on another station.  Wireless connectivity enables adding of Metadata to shot log files and finally instant access to the footage for instant editing.

Yes, that is DEFINITELY a cool and useful option, and brings the overall package price down considerably as well.

Red One: the form factor isn't finalized, but if you've been keeping track you can see the evolution of the design is stabilizing into something fairly predictable at this point. They've said all along that the camera body weight would be under seven pounds without lens, battery, or recording media. Not too shabby for a 4K capable system.

It is small, kind of barrel shaped. I've yet to see a grip to try to hand hold it, and the shoulder mounting rig, the Red Rail, while expected to be of lightweight, strong, high tech materials, is kind of bulky (I've doodled with the earlier prototypes). Ah, I'll hold off on all that until accessories.

The main gist of the form factor here is that it is modular and flexible in intent - it is a fairly small camera, and you can remove EVERYTHING - even the battery, viewfinder and LCD flipout - to keep it ultra compact and portable. For the steadicam operators I've talked to at NAB & IBC, they really liked the idea of having just the camera, a lens, a battery, and the onboard RedFlash to record up to 18 minutes of 24p footage (at 4K RAW no less) on a 32GB solid state recording media (no dropped frames due to jostling of media...so does that mean when the drunken sod/cyclist/race car hits the poor cameraman, we won't see that jarring lost frame and static? Somebody go test with an HVX200...)

Need a big camera to load up? The Red Cage will give a solid foundation to mount whatever you need onto...provided those accessories have the right screw-on threads to be compatible.

WINNER: Not entirely clear. Neither form factor is finalized, and the Red One's accessories will be vital to how well it configures for particular tasks. If you need a tiny head with remote recording, to shoot in tight spaces, SI-2K has a definite advantage. If you need fully self contained recording in the smallest/lightest possible form factor, I'm betting Red One will have the edge. If you need to rig up for "heavy" shooting, Red Rail and Red Cage suggest more flexibility than SI-2K. So a nod in the direction of Red One, but this is getting into a Horses for Courses kind of a thing, without definitive overall winners unless you get into the specifics of what you want to shoot. Neither of these cameras are going to be palm cameras, that's for sure. Sticks or shoulder mount or an awkward hand carry I'd guess (until we see finalized stuff from both).

RESOLUTION & RESOLUTION OPTIONS (click on links for manufacturer's pages on these details)

SI-2K: 720p, 1080p or 2048x1152

Red One: 720p, 1080p, 1080i, 2K (windowed or scaled), 4K, 4.5K

WINNER: Strictly in terms of OPTIONS OFFERED, Red One, no arguing there. They are both single sensor CMOS CFA cameras and generate their pixels in similar fashions. So unlike pixelshift arguments, or 3 CCD vs CMOS CFA, or CCD arrays, this is a fairly apples to apples comparison.

UPDATE: Mitch Gross duly busted me on CML on this one - well yeah, there is wiggle room the way I had it phrased before. What is the quality of pixels and resolution? What is the underlying CMOS technology in each? Valid questions all, so I'm rephrasing this section. But look at the sample footage, and draw your own conclusions. Raw pixel counts do not guarantee a winner. Based on that kind of analysis, one could claim a sub-$5,000 HDV camcorder was "better" than a $65,000 Varicam, since it has more pixels on the sensor - not the case.

In terms of the actual, delivered resolution, test charts will be the ultimate answer on that. Based on published images, however, I'm leaning in Red's direction, but that is opinion, not backed up with hard data at this time. But I suspect Red's 4520x2540 will generate a sharper image than SI's 2048x1152.

Sensor Size

Both use a single sensor CMOS with a Bayer pattern. Kudos to both for using larger sensors, means better depth of field (freed from the curse of 1/3" sensors, which means the glass has to be massively better to resolve to those tiny sensors). In this price range, you get 1/2" HD sensors (Sony XDCAM HD) or 2/3" SD sensors (Panasonic's SDX900) at best. Here we start at 2/3"...and keep going up. Bigger=better in this context.

SI-2K: If I recall correctly, the SI-2K is using an off-the-shelf Altasens sensor. Ari tells me it will do a native 2048x1152 resolution. It is a 2/3" inch 16:9 sensor. It is the Altasens HD4562.

Red One: Red's Mysterium Sensor: I don't even know who makes it, and I've worked their tradeshows, twice. They are very secretive about its origination. But Jim has tauntingly named it the Mysterium, and it is a Super35mm, 24.4x13.7mm sensor. BIG. Actual usable pixels will be 4520x2540, but with 4K, 2K windowed (uses a smaller, inset area of sensor, cropping the active area down), 2K scaled (shoot 4K, scale to 2K), 1080p, 1080i, and 720p options.

WINNER: Red's Mysterium - bigger is better in this case - gives the option of matching Super 35's depth of field, which the smaller SI-2K's cannot. Note this category is merely sensor SIZE, not sensor quality.

Frame Rates:

Since neither system uses* tape, both are freed from the tyranny of the need to maintain a constant tape velocity, so both are freed from that onerous constraint when it comes to recording varying (and perhaps variable) frame rates. So both can record whatever the REST of the system allows - compression speed, data throughputs, sensor sensitivity and recharge/recycle time, etc. Since they are both digital systems, frame rates are tunable as finely as you'd like, and I'd imagine both will allow for shooting things like pi frames per second, etc. Of course, both support the fractional rates like 23.976 and 29.97.

Bravo to both for freeing themselves up to record data at whatever rate they wish, without flagged/duped frames and other cumbersome technological hacks.

*(I'm going with the present tense out of convenience, although I'm well aware neither is shipping yet)

Both also appear to be using sensor windowing tricks (using less area on the sensor, a "windowed" or cropped portion of it) to boost frame rates:

SI-2K: at 1080p and 2K resolutions, up to 30 fps. At 720p, up to 72fps....but strictly progressive in all modes. Since capped at 30fps for 1080, can't derive a 60i from a 60p.

Mid edit update: Oops, not quite right, here it is from their website:

-2K Formats: 2K/23.976p, 2K/24p 2K/25p
-HD Formats: 1080/23.976, 1080/24p, 1080/25p, 1080/29.97, 1080/30p, 720p (variable)
-72fps/720P for slow-Motion Special effects
-Overcranking and undercranking for special effects (12~72fps)

...and I'm guessing 12-30 fps for 1080/2K res, 720p only for 30-72 fps based on the above

Red One: a little more complicated - if using the full sensor area, up to 60 fps (can generate 720p, 1080p, 1080i, 2K, 4K & 4.5K with this)...but only on certain recording media. If using the windowed (cropped center section of sensor, applicable for using Super16 & B4 mount lenses), can go up to 120 fps....but only up to 2K resolution, and only on the RED-RAID. Onboard recording capped at...ah, hell, I dunno, they've been talking about 4K REDCODE compressed recording onboard, and clearly this chart isn't updated. But while the CAMERA is capable of up to 60fps at 2540p and 120 fps at 2K (windowed/cropped), it doesn't appear that the onboard recording devices will necessarily be able to record those frame rates. All squares are rectangles but not all rectangles are squares, YMMV, so I'll keep up on this and you folks pay attention when you plan your shots once these come out.

UPDATE: Graeme posted on a board today about hardware compression in camera limiting frame rates for Redcode RAW:

REDCODE frame / sizes are constrained by the max bandwidth through the compression chips. Once we know how they behave in practice, we can re-evaluate those limits, but for now, the maximums are 4k 30p and 2k RGB 60p.

WINNER: Red One - higher frame rates at higher resolutions consistently, even if the limitations are squirellier/more complex. As the SI-2K is focused on being strictly a d-cinema camera, they didn't include 1080i. Again, the Red offers the same but more, including the broadcast necessary 1080i formats.

UPDATE: Actually, not quite as clean cut after all - the Red camera is capable of 60 fps in 4K mode, and 120 fps in windowed 2K, but to record that fast, you'll need REDRAID or REDRAM or something as yet undiscussed. 30fps max onboard for 4K RAW. Want 60fps? You can shoot full aperture with your Super35mm lens, but you'll have to record at 60fps 2K RGB - it will be downsampled, but still full range - you've only converted RAW to RGB and scaled it down. For 120 FPS with the windowed 2K, again you'll need something more than the REDFLASH or REDRIVE - 60fps seems to be the max. So onboard with the everyday recording media it looks like the SI-2K will go faster (72 vs 60 fps), but the resolution will be lower (720p vs. 2K derived from 4K).

Recording Modes

This can encompass several things, so I'll touch on them one at a time, some of which we've already covered:

-physical configuration modes - covered above in form factor - a bit fuzzy, but seems to be a nod towards Red more often than not, but YMMV depending on your shoot's needs.

-Frame size/frame rate modes - covered above in Frame Rates, Red One gets the nod.

Recording modes: both record to 10 bit, full raster, proprietary wavelet based RAW codecs recorded onto a 2.5" SATA drive with standard computer connections. UPDATE: Redcode is 12 bit I've just learned, go read this and this

SI-2K: But when it comes to recording modes, now we're into new territory, where the nature of the two cameras starts to come out. The magic deal with the SI-2K is that it is a computer made into a camera. This has many advantages - coolio UI, ease of adding new features, ease of upgradeability for certain things, etc. But...since it is a small form factor computer, it lacks some things you'd otherwise expect on a video camera, such as traditional video, rather than just computer outputs. This all leads me to say that there are really just a few possible ways to record the output of this camera, and all are data centric:

1.) Most folks, most of the time, will probably record 2K or 1080p as Cineform RAW onto the intended USB 2.0 2.5" hard drive as Cineform RAW. From there it can be directly and natively edited in Premiere Pro with Prospect 2K (or Prospect HD for some). This can occur whether the MINI head is directly attached to the DVR as a camcorder unit, or whether the head is removed and the DVR remoted via GigE cable. Easy as pie to transfer to your system, no fancy card required, just plug into USB 2.0 port and copy over data - pretty much any vaguely modern laptop will work just fine for transfer. And you can hold up to about 4 hours of 1080p footage on a 160 GB drive. Wow.

2.) Some will want maximum quality, such as perhaps for best possible green/bluescreen shots for compositing. Then they'll just record the uncompressed RAW to a file (I'm presuming there's an uncompressed RAW codec option as well). And is that uncompressed RAW editable on the Premiere Pro/Prospect HD timeline?
UPDATE Ari Pressler of SI:

UNCOMPRESSED = 12-BIT LINEAR!  We record each clip into a single RAW file, which anyone can read and access the data or segment into a sequence of DNG frames. (Mike note: After Effects can read these in now)

3.) If you really, really wanted to, you could convert the DVI output from the DVR to HD-SDI with an external converter box, BUT you'd be limited to 8 bits/channel. Anyway, with that HD-SDI you could go to a traditional deck, DDR, broadcast truck, whatevah...but what a sucky option compared to the price/performance/quality you'd get with native RAW recording.

The saving grace here is that SI is going for the modern take on imagery - the first thing you're going to want to do once you have an image, especially if for anything other than live coverage, is put it on a computer and edit it. You're getting it there immediately. The plus is great quality recording, immediacy and access without an expensive deck, but the downside is data is awkward and expensive and time consuming to archive, rather than "I dunno, throw that used tape up on the shelf there."

(Red is in the same boat too - data acquired to be backed up).
UPDATE:
Ari of SI wrote in to point out:

SI-MAG - The removable mag for the SI-2K will ship with 160GB.  Additional empty mags will be less than $100 and you can put in your own 2.5" HDD for about $40/Hr of material (80GB hdd are now only $70!!).  Why not put these on the shelves just like tape?


Anyway, those are your options as far as I see it.

Red One: Again, the computer w/lens vs. flexible camera pops out as we compare the two.

Recording options on Red One:

-it has single and dual link HD-SDI connections for those that want to hook into existing broadcast infrastructure. That right there may be a make or break for many potential purchasers. And yes, those outputs can be "normalized" and not look RAW. You can do 4:2:2, 4:4:4, or 4:4:4 log (think Panalog or Viper Filmstream) modes. These only work up to 2Kx1080 (I think, based on this chart), however - beyond that, there IS no single/double cable standard (Sony's 4K projector uses a fistfull of HD-SDI connections...who wants that coming off a camera?). From there you can run to any HD-SDI equipped deck, DDR, whatevah - you're covered. Dual link HD-SDI to HDCAM SR or S.two or Codex box? Yep, all options.

-RED-RAID - for the full-on, maximum, no compromises quality, there will be a product called RED-RAID. Size, capacity, recording duration, price, ship date, all unknown. But it will record ANY output the camera is capable of, up to and including 60fps 4520x2540 uncompressed 12 bits/channel. This is pretty much spec-ware as far as I can tell - they haven't announced/said peep about it publicly, and the ship date is sure to be after the camera....how far after is the potentially ominous question. But the max datarate the camera could generate is over 900 MB/sec - which would require no less than about 15 fast SATA hard drives, more like 25-30+. So yeah, you're not running around with that many drives, plus RAID controllers and batteries for the whole gob on your back. Think mini-fridge. Smallest unit I've seen capable of that was based on 40 (yes four zero) 2.5" hard drives, was about the size of a G5/Mac Pro tower.

RED-RAM - again, no price/size/capacity/ship date/etc. info, the idea on this one is it is a high speed, uncompressed, small-enough-to-carry, solid state recording module so that you could record uncompresssed and be self contained. If you've heard of the Venom recorder for the Viper, same concept, but hopefully substantially less pricey - I think that thing is north of $50K, and records about 8 minutes or so if my dim recollections are correct.

RED-FLASH - for compressed recording only, up to 4K (not 4.5K), using REDCODE RAW or REDCODE RGB. Solid state, so no risk of dropped frames due to g-shocks throwing off the heads in a drive. Dunno if the rest of the camera could hold up to the abuse (or if I'd want to test this theory on a camera I owned), but you could record to this kind of media on a paint shaker and not drop a frame in theory - so feel free to run around and not worry about dropped frames, just dropped cameras. Higher cost/GB, so lower capacity per dollar. They've stated they'll start at 32 GB, adding 64 and 128 GB capacity as well over time, starting around $1000. At the data rates for 4K @ 24fps discussed at IBC, that's about 18 minutes of 4K RAW 4:4:4 10 bit log REDCODE RAW per 32GB cartridge that fits inside the camera body.

RED-DRIVE, aka Digital Magazine - OK, this is what 90+% of the folks will use 90+% of the time I'll bet in "normal" shooting situations - based on a 2.5" SATA hard drive, with eSATA, USB 2.0, FireWire 400 & 800 connectors. If you don't have one of those connectors, time for a new computer. Anyway, they've talked about these being based on current 2.5" hard drives of 40-160 GB capacity (and since IBC there are now 4200rpm 200 GB models). They've publicly stated (with usual "all specs subject to change" caveat) that these will start BELOW $1000. So worst case scenario, for $1000 you'd get 22 or so minutes of 4K, 24p footage. Copy it off to laptop or external drive connected to laptop, then keep shooting.
This is the same basic principle as the SI-2K.

WINNER: While they have their commonalities (10 or 12 bit, wavelet based, proprietary codecs to 2.5" drives onboard, or 12 bit uncompressed recording), Red offers further options beyond that, so I've got to give it the nod there. EDIT: Plus Redcode is 12 bit, an edge there as well.

UPDATE: Ari of SI pointed out:

SI-MAG - The removable mag for the SI-2K will ship with 160GB.  Additional empty mags will be less than $100 and you can put in your own 2.5" HDD for about $40/Hr of material (80GB hdd are now only $70!!).  Why not put these on the shelves just like tape?


That's a very interesting question. The fact that SI is selling empty mags for you to put your own drive into is a definitely cool thing. At $70 for 2 hours of footage at best quality, think of it as the equivalent of a $35/hr tape. Red, from what I can tell, is going to sell their digital magazines as fully self contained things, not an empty shell to pop a drive into. I've heard them discuss at trade shows concerns about G-shock tolerances, and drive speeds, and other factors that will make some drives suitable and others not - so I'd guess they aren't going to have an open architecture in that sense - there won't be "empty" cases for $100. Part of this is blunt necessity - the data rate is about twice as high as the SI-2K's, and that gets into "yes it can or no it won't" territory on these smaller drives. Redcode at 4K doesn't have the safety buffer (due to higher data rate) that Cineform RAW does at ~15MB/sec. So shelving digital magazines won't be cost practical for Red. For SI, what if you pull the drive from a mag and shelve it? It's $35/hr then. Leave it in the shell? $100 shell plus $70 drive-$170 that holds two hours, so $85/hr. Would you put $35 to $85/hr tapes on the shelf, or back'em up? For ENG and some docs, that's too expensive. For a feature, maybe that's fine. So it is a possibility for SI, not really one for Red to just shelve the "shot on" disks. Shelving disks is decently stable, but with moving delicate parts, not nearly as stable an archival medium as video or data tape. But either could be copied onto cheap 3.5" drives, that's another pseudo decent backup strategy. More some other article.

AUDIO

SI-2K: only temp audio using the built-in stuff. There is lots of discussion about doing things like using USB based audio devices, but it isn't there yet, and I"m not exactly sure that they plan on it - it is a bit of a philosophy thing - they see this as a digital cinema camera, and cinema (film) cameras don't record audio onbaord, do they? I just don't like the thought of having to sync in post - whadda pain. But this may change before it ships, or even after it ships.

Red One: 4 channel uncompressed, 16 / 24 bit, 48KHz, with some kind of inputs/outputs onboard (dunno if XLR (likely mini), AES/EBU, or what at this point), with embedded audio on the HD-SDI's outbound. Rumors and hints on the boards it MIGHT be better quality than stated specs as well.

WINNER: Red One as it looks now - a different philosophy, they planned on this to act as a video as well as d-cinema camera/camcorder, so audio was planned from the beginning. Lots of high quality channels, in and out, can do outboard sound and bring it in or sync in post. Again, Red's thing about "the power of AND, not the tyranny of OR" pays off in my opinion.

OUTPUTS:
Both offer at least a digital out for at least computer screens - a great way to have inexpensive (if not entirely accurate) monitoring, but is a GREAT solution for determining proper focus - a 23" 1920x1200 inexpensive LCD doesn't give a false sense of security, it lets you see pixel for pixel of a 1080p signal.

SI-2K: On camera, you've got the touchscreen viewfinder and a DVI output. That DVI can drive a 1920x1200 pixel monitor (at least). These are, essentially, the VGA (touchscreen) and DVI outputs of a twin-head computer display. No electronic viewfinder offered - you're using the touchscreen to frame when portable.

The good thing about the DVI out is that pretty much ANY "normal and reasonable" computer LCD panel will work, since it IS being connected to a WinXP computer. The catch might be how to get it to drive at the resolution and refresh rate of your choice...what's the UI for that? Can it sufficiently auto-detect? I'm not saying it is a problem, I just don't know.

Red One: On camera, you've got:

-a probably (no specs yet) non-touchscreen LCD panel, about 3.5" it would appear

-a (currently slated to be) optional electronic viewfinder that will be 720p, price and availability TBD

-single and dual link HD-SDI - 4:2:2, 4:4:4 linear (vid gamma in Stu-speak), 4:4:4 log (again think Panalog or Viper Filmstream if you're familiar with those)

-HDMI output - can plug into a consumer HDTV for inexpensive monitoring of audio and video...but since HDMI is a superset of DVI, you can also connect it to a computer LCD flat panel as well, although the exact limitations on pixel dimensions and refresh rates supported might get tricky.

-some kind of audio I/O, unknown what exactly (see above)

-unknown whether they'll have a 9-pin deck control cable for using the camera as a virtual deck, I highly, Highly, HIGHLY encourage them to do so (Ted/Jim/Stuart, you reading this?). This will make it possible to use the camera as a deck for any setup that isn't otherwise compatible for ingest with the camera, or for realtime ingest when in a hurry (dailies, news, etc.)

...so that ought to take care of ALMOST any need for connectivity

Neither have high definition analog component outputs, which is what many would like to use to connect to color accurate CRTs. While you can get HD-SDI boards to connect to color accurate CRTS, those boards are usually in the $3000-$4000 dollar range...ouch. You can convert HD-SDI pretty readily (and probably DVI to component if you had to, but that's a funky trick and probably not cheap)

LENS OPTIONS:

Now, I'm really a post guy - my knowledge/understanding of cameras starts at the optical block and goes back from there into the camera. Lenses - I have merely a basic understanding of day to day lens stuff. I know shallow depth of field is considered preferable; I know bigger sensors mean lower tolerances for the glass to achieve a given level of quality (therefore big sensor=good, small sensor=not as good, since requires MUCH tighter/more accurate tolerances to achieve the same resolved image quality) - stuff like that.

As for mount technology, I don't know a lot. I know PL mount is what is used for traditional film cameras, so interoperability there is a big plus. Both of these cameras can take a PL mount camera, but since the SI-2K has a roughly Super16 sized sensor, 35mm lenses will fit but have a roughly 1/2.5 magnification effect. Less than optimal, since the optical functionality isn't as expected. But it will also take F & C mount lenses, good for the low budget set.

The Red One will take Super35, 35, and Super16 PL mount lenses by as shipped/by default, and there will be an optional B4 mount (the bolt circle on the front of the camera actually does something - take off the mount, bolt on another for a different lens format). There has been discussion on the boards about using SLR still camera lenses, but I don't know what mounts they have or haven't committed to yet. Since I feel ill equipped to properly judge and weigh all these factors, I'll just say this is the information I have at this time, judge for yourself. As both get closer to shipping and All Is Made Clear, this'll be easier to pick.

UPDATE: Ari Pressler gave me some more details on the SI-2K lens options:

We have C-mount and PL-Mount options (no B4 mount yet due to the fact most B4 lenses were designed for prisms not single sensor planes).  For fixed mount POV shots or a low cost 2K solution you can get a complete set of Fujinon c-mount lenses (12.5mm, 16mm, 25mm, 50mm) for under $2000.  These small form factors are great for fixed position shots or POV shots where the camera may even be in the scene .  The format is almost identical to 16mm, so everyone that has an investment in these optics will be quite happy.  35mm can be used but you get a 2.5x or more multiplier in magnification.

I've heard discussion on the boards from Red team members about still lens mounts, so there is some commitment there, but I don't know exactly how that is going to play out. You don't get follow focus gearing, those lenses aren't set up to focus the way you'd want a video/film lens to focus, so it is a hassle, but it CAN be done with varying levels of efficacy, YMMV. UPDATE: Jarred Land wrote in to point out you can get follow focus gearing for still lenses as an after market thing.

So both will have PL and still camera lens options, Red will also offer a B4 mount, but the SI camera's smaller chip will mean Super and regular 35mm lenses will have a magnification effect. Sounds like Red has matching and/or superior options here, but SI-2K still has good options.

CODECS

Both are 10 bit, variable bitrate, full raster, RGB, 4:4:4 wavelet based proprietary codecs. UPDATE: Redcode was just announced to be 12 bit, with 10 bit linear, 10 bit log, and 12 bit linear options I don't have enough information to say which is "better" at this point in time in terms of quality. As a practical matter, there will be other codec factors, tying into workflow (more on that in a minute). Both get high, HIGH praise for Doing It Right in terms of very efficient RAW codecs using wavelet technology. I know that the Red codec is definitely log encoded, I don't know whether the Cineform RAW codec is or not. Somebody clue me in. Red is going to also offer RGB codecs with 4:4:4 and 4:2:2, dunno the matching details on Cineform's codec stuff.

More Update: Ari Pressler of Silicon Imaging wrote me to say:
CineForm visually perfect codec generates ~15MB/sec for 2K.  That is what gets us the 4-hours on 160GB-HDD.

WINNER: too early to say - Cineform is out and working, Redcode is still under wraps - we've seen some demos, but not worked with it. Final image quality is the ultimate decider and Redcode isn't finalized to say. Redcode is capable of greater bit depth, which is preferable, but the final, ultimate answer is in the compressed images, and we haven't seen enough of Red to say. Visually lossless (no visible artifacts) is great, and both are claiming that, but the REAL test will be how well they hold up to aggressive color correction. Compressing for optimal human perception is not the same thing as compressing for optimal post manipulability. So even if "visually lossless" is achieved, that doesn't yield the same results in post as mathematically lossless (where every pixel exactly matches the uncompressed source).

ON CAMERA USER INTERFACE

SI-2K: I've seen it and played with it, and it is pretty damned slick, all touchscreen driven, deep menus to do things like change the name/file directory your shots will be written as/to, still store with 50% overlay for reframing shots to match (MAN THAT IS NIIIIIIIIIIIICE!), touch center of screen to zoom in for focus assist, an exposure assist mode that uses false colors to warn when clipping or almost clipping, etc. When I saw it six months ago at NAB it suffered "Designed by Windoze Programmer" syndrome, but I'm betting/hoping it has evolved since then - that was the first public demo...plus I'm a UI snob (used to work at frogdesign, a UI house, and have lots of ex-coworker friends who still do that for a living). But the functionality rocks, and the fact that it is just an application they can always add to means new features in the UI or a snap to implement.

Overall, with the touchscreen and the more computer-like interface, MUCH easier to navigate and use. A HUGE step forward for camera controllability as I see it. No more pushing little buttons 37 times to get to the menu you want at the bottom of the list (HATE that).

UPDATE: Ask and ye shall receive - Ari sent this over, since he's sent a picture, you'll be spared my 1000 (more) words:

: )


click to jump directly to a larger view, Control click to select to open in new window


Red One: we've seen a few slides in presentations that haven't been made public. They are promising all kinds of coolness, but we haven't seen it. Also, as an embedded UI, it is inherently more difficult to update or especially augment than a Windows app. We have seen a dial on the back of the camera, somewhat enigmatically iPod scroll-wheel like, we'll have to see how that ends up being implemented. There's a small screen on the back of the camera (last I saw) for strictly UI stuff I'd guess, but we have yet to see a "real" screen on it, for it, or from it.

WINNER: For now, the SI-2K...because

a.) they have one, and

b.) the easier future upgradeability

Of course, we'll have to see what Red does. No telling when we'll see anything of the UI. I've heartily encouraged them to break free of the 1980s DOS design ethos that most camera menus work with, we'll see what they come up with.

IMAGE QUALITY

This is where things can get mightily subjective, and so I'm going to simply direct you to the resources, and you folks tell me what you think. Objectively, the Red can achieve a shallower depth of field, and that is considered good, and should be able to generate a sharper image with the higher resolution sensor. But, again, since SI is further along in their development, they have a MUCH better range of test footage that they are showing off - outdoor shots as well as indoors, and most importantly, stuff shot in the field by non-vendor operators. Both manufacturers have stills and videos posted, but SI has more of them in a greater variety, if only because they are closer to done.

SI-2k:

SI-2K video footage gallery - 20 shots, in up to 1080p resolution, Windows Media 9 and MPEG-2 encoded, also includes some Cineon/DNG stills at full res.

Silicon Imaging Stills Gallery - unfortunately 1/2 res only JPEGs - a decent general overview but the over link gives better footage for evaluation. 5 shots with a variety of scenes - closeups, indoors, outdoors, people, landscapes, etc.

Red One:

RED / Gallery - of stills. JPEGs from uncompressed and REDCODE RAW encoded source, at 4K, 2K, and 1K resolutions (unfortunately not all images at all resolutions, it is a mix, one size per image). At present, only four images, but others have been posted in the forums on DVXuser.com and DVInfo.net in the past...just dig around and search for posts from Jim Jannard with attachments.

RED / Gallery - two Quicktime movies to download as .torrent (BitTorrent) files. Indoors, controlled lighting, not full res (not like anybody can monitor a 4K movie 1:1, anyway). As Red isn't as far along in development, they've been slow and cautious about what they show publicly, revealing images and footage that is nice, but doesn't REALLY test the sensor - no publicly shown/posted outdoors footage, for instance. I'd hope/expect this to change after the November 14th public screening - maybe they'll post some of that footage on the web.

The optimists say it is a marketing thing to keep interest up, tease/trickle out new footage, the pessimists say they are hiding their flaws by not showing footage that truly tests the dynamic range of the camera.

WINNER: Well, no official winner here. SI has a very wide range of sample footage, some of which shot in challenging situations (outdoors, for instance, which Red hasn't shared yet), while I've personally seen 4K Red footage projected (of all controlled indoor studio stuff) and it was VERY sharp and impressive, and a few stills that haven't been publicly shown yet.

Unofficially, my may-well-be-too-biased opinion is that I expect the Red One to be a more capable camera when it ships, in terms of dynamic range, resolution, and favorable color reproduction, but there is not publicly posted footage to back up that supposition, so we'll call it my I've-spent-lotsa-time-with-Red-and-not-Silicon-Imaging hunch for now. If I'd spent more time with SI, I'd probably want to lean more in that direction.

So officially, no declared winner. But SI gets the prize for putting forth the better range of sample footage as of this week, and gets the "money where your mouth is" nod...as of now.

Of course, the truer test will be to get these out in public and let us play with them, or better yet compare them side by side.

UPDATE: Just talking of Red (by itself, no comparison/mention of SI-2K), David Stump (ASC) likes what he sees of the Red One when he shot with it for a day doing greenscreen tests.

WORKFLOW IN THE FIELD

While the Red One has more options (see Recording Modes above), I think the majority of users of BOTH cameras will record their data in the same way - to the 2.5" drive modules using the respective native, proprietary codecs, 10 (or 12 with Red) bit wavelet compressed or 12 bit uncompressed, full raster, etc.

That said:

SI-2K: As I mentioned in the form factor section at the beginning of this beast of an article, the SI-2K can operate as a camcorder or as a split shooting head/DVR setup. Another big, BIG thing they've announced is support for non-destructive LUTs, including 3D LUTs, that can be associated with the footage on a shot by shot basis. Go read the big long post I did the other day on this, specifically the Iridas .look file stuff, It Is Important. The ability to assign different non-destructive LUTs for previewing and recording is SERIOUSLY nice, and a HUGE step forward.

UPDATE Ari Pressler of Silicon Imaging wrote in to point out an oversight on my part:

I think one very powerful configuration is missing in your analysis: The  Core 2 Notebook + MINI.
 
Core 2 Notebook ($2K) + MINI ($12.5K + Accessories)  = 2K Recording,  look editing station and HD monitoring in Component and DVI for direct projection.  If you have a fast enough disk system or enough RAM buffer you can even record 12-bit content, export to DNG (or DPX sequences) and use a fully uncompressed workflow that you already own (no additional cost)!  Plug in USB drives for additional mags for the local computer retailer on the corner.
 
I can shoot on the laptop, edit and transmit/network all from the same device!!
 
Wireless access to the recorder allows remote desktop operation or exchange of .look file information which can be generated on another station.  Wireless connectivity enables adding of Metadata to shot log files and finally instant access to the footage for instant editing.


Yes, that is DEFINITELY a cool and useful option.

Red One: The Red will have a lot of configurations - either gussied up big with all kinds of options hung off the Red Cage or Red Rail systems, or stripped down for Steadicam or handheld work. I haven't heard anything publicly said about LUTs, but it is a known industry issue, and they still have months before they ship.

WINNER: Now, here's where it gets dicey and verrrrrry subjective. There will be those for whom the MINI head configuration is a huge deal - Geoff Boyle is shooting a feature called the Mutant Chronicles and used the MINI head at 72fps to shoot some high speed stuff and in-the-crowd stuff. For some this will be the mission critical factor that makes it a no brainer. On the other side of the fence, the Red One doesn't get THAT small/light/portable, BUT it can be configured at around 15 pounds or less (we've guesstimated) for a self contained, battery powered, 4K Steadicam package with camera, lens, battery, RED-FLASH module, and not be tethered at all. It can also be built up big to handle all kinds of bolt-on accessories, and work with all kinds of fooferall hung off of it - because it is built for that.

You decide for your own projects.

My thought is that more often, most of the time, the overall flexibility of the Red One will get the nod for the greater variety of projects and shooting situations, disregarding the image quality of either camera. But for others, that saucer section separation will be VITAL, or the use of the head with a laptop as well.

When shooting in field to the compressed hard drive recorders, the challenges of in-field backup will almost certainly be identical - you'll need a way to verifiably copy off the data, make sure your copy is good, and ONLY THEN wipe it to return it to use. OR you'll need enough to shoot all day, then it is someone's job to take care of that task before you start shooting again, likely the next day. And if a small crew, someone who has been working all day is probably THE LAST person who should be in charge of deleting all the shoot info off the portable drives.

Suffice it to say, it's an issue of importance, and a new one in this post-tape world.

Blah - I could go on for paragraphs on all this, I'm not going to now. There's lots more to field workflow to be analyzed and discussed though.

UPDATE: OK, it's the next day, and reviewing all this I have some new thoughts at a broader level:

SI-2K: it blurs the line between production and post. With the nondestructive LUT embedding, the ability to record one and preview another, the still store with overlay capabilities of the DVR, the playback and review capabilities of the DVR, and the split head/DVR functionality. It is novel, original, and useful. They've placed their emphasis on this functionality. And since it is just an application running on upgradeable hardware, I see Silicon Imaging continuing to extend the capabilities of the DVR quickly and inexpensively. Ramped frame rate shooting with curves and graphs and triggers? Time lapse? Live grading on set? All doable stuff, they just have to write a Windows app, which is trivial in comparison to pulling that off in dedicated embedded hardware. This will be the area in which SI can potentially shine in the future - high level functionality at a relatively low cost onboard their fairly small system, leveraging that cheap, powerful commodity hardware and the underpinnings of a modern OS. It seems they've focused on building a camera for the post centric pixel crowd like me.

Red One: the Mysterium sensor is really the secret sauce here - the images so far look really good, the resolution and frame rate stuff is amazing. Without it, this would be a much simpler comparison. The real strength of the Red One, beyond that, is flexibility and integration. More emphasis has been placed on interoperability with existing infrastructure, and the shootability of this camera - the Red Rail and Red Cage show this thinking, as does the complete Lego-like modularity of everything. Plus, we don't even know all the capabilities yet - they are further from completion, and haven't revealed all the details yet on their camera. With their clearly larger development budget, they are managing to pull off a lot of things in compact, low power custom silicon that Silicon Imaging is doing on bulkier commodity hardware. Overall, it is a better integrated camera for shooting, and acts more like a camera is traditionally expected to, just with more and better specs than existing cameras have had, with new goodies like tapeless data centric workflow, etc.

WINNER: Will depend on your needs. Personally I like what Red is doing and I'm thinking it will appeal to a broader range of users, but your own needs will vary - what is popular doesn't matter; what works best for you does.

WORKFLOW IN POST

OK, you've shot your footage - now what?

Keeping it simple for the moment, let's pretend you recorded to the 2.5" hard drive solution for either camera.

SI-2K: you copy of the data to your work disk, and if you're running Premiere Pro with Prospect 2K or Prospect HD, you're golden. If you've embedded LUTs, those come along for the ride as well, although exactly how all that works hasn't been revealed or demo'd yet. You can play back, edit, cross dissolve, color correct, etc. in REAL TIME on a sufficiently pimped, up to spec system. It will also graceful degrade under pressure - if it can't pull off the effect in full res & full speed, it'll drop res and/or frame rate to achieve realtime playback. Very nice.

At PRESENT, they have a working, shipping editing solution for HD - Cineform provides that as an add-on to Premiere Pro.

BUT...if you're wanting to edit in Final Cut Pro of Avid...you have to transcode. A time consuming pain. (Redcine, same boat.)

Their plan is to have a QuickTime codec, and it sounds like the goal is to have realtime performance, at least for playback if not effects, within Final Cut Pro.

It is also unclear what the exact workflow will be to get from Cineform RAW compressed into Iridas' Speedgrade. Time will tell.

The plan has long been to get a Mac QuickTime codec implemented, but even if they do that, it isn't a complete solution, or it is at least a limited one.

Even with a Mac QuickTime codec, there will still be issues:

-Cineform RAW decodes to an RGB 10 bit 4:4:4 codec. At present, only Apple generated, 8 bit, 4:2:2 Y'CbCr (aka YUV) support RT Extreme to do things like realtime cross dissolves and color correction. Even if this third party, 10 bit, 4:4:4 codec WERE supported by RT Extreme, FCP at present can only do 8 bit RGB or 10 bit 4:2:2, but not 10 bit 4:4:4 RGB. All barriers to smooth integration, until/unless the major rewrite of FCP I've been hoping for is underway...and since Cineform has been talking about slim resources to help them port even the codec to Mac OS X, I think odds at this time are slim they will get prioritized to get native support in FCP.

By the way, Redcode faces the exact same challenges as well.

So therefore, for efficient editing, some kind of offline may well be required. Especially since the hardware requirements are HIGH for 1080p Cineform RAW editing. And that'll require an offline/online workflow, and conversion...hence transcoding. I asked how this might work, and they suggested After Effects to convert the shots. Which works, up to a point (you lose your timecode, and possibly your metadata, either ditching it undersirably, or baking it in, potentially undersirably). See the above workflow linked article for a link on bludgeoning After Effects to work in batch convert mode.

BUT...I'm talking about the negatives here - if you have or can live with Premiere Pro, with Prospect 2K/HD, on a hoss enough system, and want to grade with Iridas, you're in a great shape, ready to cut native with realtime support. But as soon as you stray from that golden path...

I'm crawling up SI's shorts on this one because they've given a lot of details to critique...which Red does not.

Red One: Now, over on the other side of the fence, Red at this time has zero public commitment from ANY vendor for native support. I know they've been talking to the Three A's from early on, but dunno how far they've gotten. Personally I did see high level Apple folks checking out the booth at NAB and IBC, so that was encouraging (I don't know the "big" Avid folks well enough to pick out their faces, nor the equivalent Adobe folks).

But what they have committed to is a lovely punt play - if you aren't going to have native editing (yet), and/or don't have the recent enough hardware or software to do it even if it were available...you've got Redcine. Redcine is an app to take in your native footage, convert it to any size you want, to any codec you have installed on your machine you could "normally" write to from any QT/WM aware app (or standard still image sequence format), and even do some simple (curves, white point, gamma, etc.) color correction on it, and save THAT metadata to be associated with the "session" if not the file.

Redcine is a Mac/PC app under development and I've just seen bits and pieces of it a month ago, and I've no idea how fast or slow it'll be to convert the footage (EDIT - yes I do - read this, it'll definitely be GPU accelerated and probably multi-processor as well), but the idea is a solid one - if you have that codec installed, or you need a DPX/Cineon/JP2K/TIFF/PSD sequence, you can get it there in a timely but not instantaneous fashion.

Redcine will work on either compressed or uncompressed footage from the Red camera.

In theory, you could also play live out of the camera over the HD-SDI into your system to capture, but that only works for HD & possibly 2Kx1080 modes at best, with some trickiness about maintaining timecode, etc.

OK, enough on this, moving on.

WINNER: assuming both were shipping and working the way they say they will...something of a muddled tie with no clear winner. Cineform already has a working solution deployed, but is limited to the #3 NLE at this time. Exactly how a future Mac codec will work in FCP is TBD, and Avid will definitely require transcoding. Red has zero committed native NLE support at this time, will also have a codec for Macs that will face the exact same challenges as Cineform in terms of the 10 bit RGB/YUV/RT muddle, but they do have the Redcine software to convert to any codec on any system. Really great functionality on a less preferred system, or guaranteed (but time consuming) conversion for everybody on everything...no clear winner there. Depends on your individual needs, and one could reasonably score it either way I think.

PRICE

SI-2K: MINI head alone, $12,500. Requires a computer to record onto, useless by itself. Want the DVR too? Then the DVR+MINI combo is $20,000. Want Prospect HD to edit with in Premiere Pro? Then it is $22,000. (And no, you can't run Prospect HD on the DVR.) The good news is that if you have an existing computer that can run the software and you can live tethered, $12,500 gets you in the door. Then you need lenses. But with lenses, you're ready to go, and prepared to edit (still need that computer & Premiere Pro) for $22K. But don't forget just buying the head and recording to your own laptop/desktop as well.

Red One: Just the body and a battery or two (and as of info I've got from IBC, no viewfinder, just LCD), that comes with no recording media or lens, $17,500. Recording media starts at or under $1000 a pop, for 32 GB solid state or 40+ GB of hard drive based storage. Want to shoulder mount? Pony up for the Red Rail system, price unknown, but in line with the rest of the accessories.

The only other prices known are for the lenses - $4995 for the f.28 300mm prime (roughly concurrent ship date with the Red One), or $9500 for the f2.8 18-85mm zoom (expected end of 2007).

So really, NOT including lenses, the SI-2K is around $22K plus "some more" for additional storage, price unknown, capacity unknown.

A similarly "ready" Red One would be $17,500 for the body, plus $1000ish for one RED-DRIVE.

Both need lenses and further accessories from there, you're dreaming of Pegahorns (horse+wings+unihorn) if you think you're done at that point. What accessories will you need? At least one lens, probably more. More recording media, amount varies. A good set of sticks. Beyond that, depends on you and your needs. Matte boxes? Follow focus? Shoulder mount? Stack of batteries? It depends.

This leads me into thoughts on all this - having spent at least a decent amount of time talking to both teams, it is clear that they both want to utilize the latest recording possibilities (RAW, data-centric, wavelet, 10+ bits, etc.), with the best large sensor they could reasonably find, and make a good camera to shoot movies with. Both teams realized that a camera was nothing without a post workflow, and so they set out to partner up with the vendors and/or make software solutions to ensure the footage could be ingested, edited, and output.

And that, very roughly speaking, based on what I know, is what SI focused on - to make the camera, using the insight of basing it on existing high powered, inexpensive commodity hardware as a quick way to get a lot of power in a short amount of time and have great flexibility to implement new capabilities.

Over in the Red camp, however, the ambitions seemed to be a little broader - Red is going to also have their own lenses and accessories, and invite others to come and make compatible accessories as well. Everywhere there was an existing standard that worked for their needs, they hewed to it. Everywhere there wasn't something good enough to meet their goals, they invented it. So the camera is pretty backwards compatible - the HD-SDIs, the audio interfaces, etc. But it also reaches forward with new capabilities as well. And to interact with more of the shooting world, they focused beyond just the camera to work in a wide range of shooting situations (little camera for minimum size/weight or big camera for a stable mounting platform?).

WINNER: the prices are close once storage is factored in (esp. with the ambiguous pricing/bundling of Redcine), the feature sets are different, but I personally feel the Red offers better bang for the probably lesser buck. The split head functionality, the added functionality of the DVR (which is great) I feel is outweighed by the superior shooting/working feature set of the Red One. Feel free to disagree with me on this, but that's how I see it.

Financial Backing & Partnerships

There's also some other factors - both cameras are from startups (EDIT: not true, Red is a startup from a billionaire, Silicon Imaging has been in business for 5 years - so both new ENTRANTS, but neither really a startup in the traditional, from-the-garage sense), but Red is clearly the better funded and backed entity. The market will have to prove whether customers are comfortable with either new startup to trust their money with. Red started taking pre-orders at NAB, and closed pre-orders at the end of October with over 1000 pre-orders for the Red One camera. Silicon Imaging is taking a more conservative approach and has not taken pre-orders as yet. Red's management and backing clearly gives them substantial lead here. While the SI-2K is a great idea, the team behind Red is more experienced, better connected, and better funded, and there were LOTS of meetings at NAB with vendors wanting to get into the Red accessories market. I don't have similar information about Silicon Imaging's backing.

Red has assembled an impressive team of industry veterans to run their program; I don't know enough about SI's team to compare so I'll leave it at that.

Both companies have established partnerships with other vendors - Silicon Imaging is further along with working software, and partners with Adobe very tightly through Cineform (another partner of SI), and Iridas recently got on board as well.

Red has announced they'll be developing products with AJA, and seems to have a close relationship with Assimilate, maker of SCRATCH (they were in the Red booth at NAB and were credited for color correcting the 4K footage at IBC), but no formal announcement has been made.

But the money question - clearly Red is massively better backed financially than Silicon Imaging...and that matters in business. Red is being taken more seriously from what I can tell, and will likely outsell SI by a significant margin. Again, this doesn't matter as much if you want one for your project, but does matter in terms of whether people will trust their money to a company for a product - will they be around to support it in a few years? Knowing more about the Red team, I have greater confidence in them, but I don't know as much about Silicon Imaging so that isn't an entirely fair comparison.

IN CLOSING

Now, I'm open to being off base here, but based on my exposure/experience I think the Red guys have it here. Superior specifications (if they can deliver, looks good so far), a broader reach & concept of execution, small enough to be nimble, ambitious of scope, great team, and enough money to get done in custom hardware development what they needed, and especially find (wherever they did) that Mysterium sensor (the true secret sauce of the Red One, without it it'd be a much closer comparison to the SI-2K).

I think Silicon Imaging has a great concept and was well timed in their goal to be a revolutionary new camera and disrupt the existing players...until Red got in the game.

SI's advantage (or niche) will be:

-first to market advantage by a few months, lets them build traction
-ready to roll with realtime support for Premiere Pro editors with a Prospect 2K/HD capable system
-available, known, and working while Red isn't
-and the biggie - that detachable head functionality - for some this will be The New Way - to let an operator work up front, but to be able to instantly and nonlinearly play back on set, test grade, etc. right there on set, practically live (or literally so). For certain shoots, this'll be great.
-the tiny head functionality - will let that setup go places a Red won't/can't fit, and will be a deciding factor for some projects and/or shots

Red's advantages, if they meet their stated goals and the specs don't change, will be:

-I think and expect better image quality, but I'm open to being reasonably disagreed with - need to see more outdoor footage of similar nature and compare/contrast between the two. People see color/contrast first, THEN resolution. I hate to pick favorites when I'm involved in the race, but that's how I think it is going to go down - the 4K stuff was sooo clean, I think dynamic range is the last area not ably proven - show us some bright daylight shots please
-more acceptance from higher end shooters with the readiness for accessories
-higher resolution
-more frame rate options
-more recording/shooting options & modes
-better depth of field capabilities
-easier integration into a wider range of systems - while PPro/ProspectHD is great, that's a smallllll subset of the market. If you're a shooter, not likely to be able to hand over a FireWire drive and say "Here you go, you're ready." to a client. With Redcine (admittedly, price unknown, don't even know if bundled with camera), there's a provable solution, even if you have to convert it for your clients yourself (yeah could AE convert, but again, a known expense and a production in and of itself).
-overall, just MORE, and more flexible choices, based on a broader vision of capabilities. That quote about "the power of AND, the tyranny of OR" says a lot

It is just a better thought out solution from what I can tell and have experienced. The SI is a glorious "Hey, what if!?" - well executed, but the enabling thought - "Build it around a computer, use off the shelf high powered commodity tech" is also a limitation that holds it back in some ways (while gloriously enabling it in others). The sensor is also a limitation - it is good...but not great. It is great for the price point...but may not be good enough as compared to the new competition on its heels.

Red has, however, been riding the ragged edge of "Hey, lookit me!" and it has garnered them some ill will from those too often burned by claims of Sliced Bread 2.0. If they should fail to deliver, or be noticeably late, or back off substantially or crucially on any of their specs, there will be a mighty hue and cry raised from the industry. A dangerous game. But...they've been delivering so far and it is looking good.

OK, it's after 1am, I've been writing for over 5 hours I think on this one (7806 words so far).

UPDATE WRITTEN NEXT MORNING: In the end, I'm starting to see these two cameras as reflecting very different philosophies, that will appeal to different users.

SI-2K: for those working on a set, that want to potentially split the DVR & head to have instant live playback from a desk on set, or need to get the shooting head in truly tiny spaces, and are happy with the image quality, this will have a lot of appeal, probably to folks more used to computers and post workflows. It is a modular form factor, with the ability to split in two as the major feature in that area. The touchscreen can be removed and relocated, but there does not appear to be much thought (as yet in last design I saw) given to mounting other accessories on the camera body. But we'll have to wait and see the final design to judge that.

Red One: for more traditional shooters, out in the field, that are more focused in shooting requirements - resolution, frame rates, dynamic range, and especially interoperability with accessories, and just a general flexibility due to the kind of modularity they've decided to go with - instead of the main body split, the smallest "atom" of the camera is the body itself, with viewfinder, LCD panel, handles, shoulder mount kit, etc. all removable and optional. The 5 different recording options, etc...it appears at present to be a better thought out whole, more of a system instead of just a camcorder. (Not that touchscreen etc. can't be removed from SI-2K either).

Overall, there's still tons I haven't addressed here yet. Check out the:

-tech specs:Silicon Imaging SI-2K and Red One

-workflow: SI-2K and Red One

-image galleries: SI-2K and Red One

Again, this is all just my opinion, and a 5 hour non-stop first draft at that with some touchups the next day. I'd like to have a balanced piece out there, so chime in with your own thoughts. I'm friendly with both teams, but obviously have a closer connection with the Red folks, so I know their DNA better. Chime in in the comments section and I'll integrate the valid points over time as I can into the text.

The good news is that these two new cameras will be coming to market next year if all goes as planned. In the end, it is surprising how much they do have in common:

-modular/configurable form factor, moreso than regular camcorders
-data centric, tapeless recording
-full raster, 10 (now Redcode is 10 or 12) bit wavelet based proprietary RAW codecs to 2.5" hard drives
-12 bit RAW uncompressed recording as well
-record to small form factor disks, then copy to a computer/other hard drive
-emphasis on the post solution with a desire for widespread native NLE support
-under $20,000
-variable frame rates
-720, 1080, & 2K resolutions all from same camera
-windowed/cropped sensor mode for lower resolution, smaller lens coverage, but higher frame rates
-higher than video frame rates
-progressive imaging
-multiple lens mounts available
-what else?

...and that means as filmmakers, we'll all benefit, not just from the particulars of these two cameras - and if they are successful, what it will force the majors to respond with?

-mike

OK, just so everybody's clear - I've worked two tradeshows with Red, I've been in meetings with them, under NDA etc.. I'm trying to be as fair as possible to both parties - am I missing any points, any categories? Chime away with the Comments link below if you think I'm off base or missing anything here.

UPDATES, ROUND 1:

-I compare the non-shipping Redcine to the currently shipping Cineform 2K, and make comments about no Final Cut support - this is temporal imbalance! Cineform is working on a Mac codec, will be able to edit in FCP, but FCP has problems/limits with 10b444 footage - Redcode will have exact same issues.

-I failed to mention the wireless capabilities of the SI-2K - I don't understand how exactly it is used, but will incorporate once I do

-BOTH cameras have upgradeable sensors - the SI-2K is an external unit and hence has more flexibility in its upgradeability - it can be whatever shape will still attach. The Red Mysterium sensor, if they come out with an upgrade for it, will have to fit the size/shape/power/thermal constraints of the inside of the current camera - a tougher engineering challenge. But both are, in theory, quite upgradeable, SI just seems like it would be easier. If they got a new sensor on par with the Mysterium, it would be a much more interesting comparison.

Links to products mentioned:

Red / RED ONE

Silicon Imaging Products

CineForm RAW Technology

Adobe Premiere

Apple - Final Cut Studio

UPDATE ROUND 2

Ari Pressler of Silicon Imaging took the time to read through this beast, and I've folded his comments into the text of the article. Significant areas of change:

1.) the ability to use the MINI head with your own laptop (not just the DVR) for capture, control, playback, look editing, monitoring, etc.

2.) more info on lens mounts

3.) Recording modes and digital magazines

4.) Data rate info on codecs

5.) screenshot of the UI

-------

also, graphic at top of article - I coulda built it, but I'm lazy, I blatantly lifted it from DVguru.com, so thanks and apologies to those guys (but Teh Luv for the link guys, thanks!)

----

Also, there's been some new developments concerning Redcine, Redcode, Cineform vs Redcode, etc., in two articles here and here

-mike

Labels: Red, SI-2K