http://www.captain3d.com/temp/cml/cml_volume.html

Also, in my previous post, I threatened to describe ways to actually fix “flat/cardboarded” shots after the stereo camera has captured them.  This is not simply an issue of translating the eyes relative to each other (which just moves the apparent convergence point).  Cardboarding is a function of a problem with what’s known as the depth budget.  To oversimplify, the depth budget is the maximum amount of behind-the-screen-depth, plus the maximum amount of in-front-of-the-screen depth.  Sometimes this is expressed in percentage of screen width, aspect ratio, or in other cases in raw pixels of disparity between left and right eye images.

To change the depth budget itself, one must literally “reproject” the eyes from different camera positions.  In the CG World, this is relatively easy, you just “move” your virtual cameras around.  In the real world, after the stereo imagery has been captured, this is a non-trivial problem, because there is no “virtual” camera to move around after the fact.  Literally, what we’re talking about is changing the apparent interaxial distance of the two cameras AFTER the take.

Clyde DeSousa has a very clever trick in his post on realvision.ae: this approach allows one to reduce apparent interaxial, and ultimately the trick in this approach is that it uses existing plugins to calculate tween images.  This is also useful for generating multiview images for autostereo displays.  He also has an excellent video channel on YouTube, a couple of his videos show this process in action:

This works in the unfortunately common instance that too much camera interaxial distance was used for the type of shot and focal length of the lens.  In this case one simply needs to reduce the interaxial, and tweening is a good process to do so, as long as the tweening/optical flow engine driving it is extremely high quality and does not leave visible artifacts.  An example of such an artifact (and Clyde duitifully calls it out) is the stretching distortion seen at the top of the example shots he shows.

However, cardboarding is a circumstance where simply reducing the interaxial distance of the virtual camera does no good — in fact, one way to fix this would be to Increase it.  A better way would be to simulate the camera being moved closer to the subject concommitantly with a change in the apparent focal length of the camera lens system(s).  This requires some more work.  There are tools that can do this — Occula and Mistika are examples.  However, as a practical matter, they have fundamental limits to the severity of impairments they can repair.  I’ve heard it said very matter-of-fact in Hollywood post-production circles that if a shot’s depth budget is overshot by more than 1.5% or so of screen width, the shot is not fixable.  Further, if the shot is cardboarded, it’s not fixable, period, end-of-story.

However, it IS possible to fix these shots.  That’s for the next post.

What’s most impressive is that the article captures the interplay of camera lens’ focal lengths, the seating position of the viewer in the theater, and volumetric perception — and does so visually. There are some rather large quickTime videos embedded in the page, I would urge patience (load the page, have a Martini, and come back). This is excellent training material for would-be stereographers. I think this is probably the finest example I’ve seen to-date that explains the issues in terms of visual intuition.

To quote Phil: “There are several large movies as I wanted to keep the anaglyphs looking clean. The page is about 150mb in total so be patient.”

http://www.captain3d.com/temp/cml/cml_volume.html

I have only one minor quibble: while stereographers shooting live-action with real, physical cameras will encounter these limitations, in an animated CG pipeline there is no reason to make your “virtual cameras” behave like “real cameras”. In a CG pipeline, the laws of physics are quite fungible, hence “multi rigs”. This quibble detracts nothing from the quality and facts of his article, however.

Nevertheless – there is a way to apply this knowledge to fix “flat/cardboarded” shots captured by real, physical 3D stereo rigs with extremely long focal lengths… a subject for my next post.

Most recently, we had the opportunity to play around with a high-speed digital cinema camera, thanks to Rufus and Chad and the other good folks at The Camera House in Los Angeles. Shooting stereoscopically at ultra high speed (say, 2000 fps) presents some real challenges, none the least of which is ensuring sync between two cameras.  There’s no such thing as HD-SDI genlock at 2000 Progressive!  We avoided that issue entirely; it’s related to my previous Kinect post, but only tangentially.  No Kinect was used nor harmed in the making of this.

I can’t say much about HOW this was done except that we did not use two Weisscam H2s, only one. I’m happy to show you the final results, though. Youtube steps on the video a bit so you can’t really see every little microdroplet of water flying at you in glorious 2K, but it’s enough to get the general idea.

Here is the RGB capture plus it’s depthmap:

The YouTube 3D version (you have to select your 3D display type):

The Red/Cyan Anaglyph version:

Among those people that follow this blog or at least find it occasionally interesting, I’m certain that 100.00 percent of you have seen the videos of the XBOX Kinect camera from Microsoft being hacked into purposes other than intended, and others like it. For those who have not been introduced to what the Kinect is doing, you can see this illustrated graphically when using a camcorder in “night-shot” mode in a dark room when the Kinect is “on”:

So, what’s really going in here is you have a very clever “structured-light” infrared shape camera. Permutations of this include depth cameras like “time-of-flight” and “flash-LIDAR” cameras, which are Much more expensive. These sorts of cameras are used for things like driving automation, critical measurement of military strike candidates (French for “measuring the height of the window the laser-guided bomb will fly into”). The Kinect is doing its measurements by projecting a very cleverly arranged dot field, whose pattern is known and predictable. Then an IR camera records the image, and the relative spacing of the dot field gives an indication of the distance from the IR emitter. In the case of Kinect, it doesn’t have to be particularly accurate — NASA isn’t using this to pick up rocks on the surface of Mars from 80 million miles away, nor is the Air Force using it for precision bombing strike selection. Kinect is used to track body movements, indoors, with limited range typical of a living room — something that doesn’t require especially precise range-to-target measurements.

Nevertheless, there are many interesting things one can use a shape camera for, even if range-to-target isn’t particularly well-calibrated. One of the first public early efforts was this:

One problem with using any of these is that Kinect does not provide what is technically called a “dense depth map” — there are lots of holes (black) where the Kinect IR sensor can’t “see”, whether due to lighting, objects being out of range, reflection, transparency, occlusion, and objects absorbing and not reflecting Infrared (which is how Kinect sees depth).  It also won’t work too well in broad-daylight, but it was never intended to.  Here’s my own example of a sparse depth map from the Kinect. Note the abundance of holes and occlusions (shown in white):

My company has been playing with this sort of thing predating Kinect by a fair bit, including using ToF (time-of-flight) cameras. We have a lot of technology to do GPU-based statistical temporal prediction, inpainting and hole-filling, in real-time — and it happens to work for Kinect, too. Here’s the previous depthmap with very rudimentary hole-filling and statistical prediction applied, and now it’s a dense depth map:

Once you have a dense, high-quality depthmap, you can render a high quality 3D video from it. You also need to be able to fill in disocclusions (a process known in the 3D film industry as “inpainting”), which we also do well, in real-time. So here is the same video in 3D using our multiview rendering engine. You will need anaglyph glasses to view it:

With a little of our magic fairy dust, the Kinect could become a truly workable real-time high-quality 3D camera, if only the Kinect RGB camera was higher quality. One interesting thing about the Kinect is that the RGB camera does not match the IR camera, so the depthmap has to be rectified to the RGB image. Even more interesting, our automatic rectification doesn’t care about the resolution or how far away the RGB camera is from the IR — in fact, the capture could come from different cameras. Like, let’s say hypothecially a 2K, or even 4K camera.

One thing is certain, people will find many, many interesting uses for Kinect and other depth cameras within a variety of industries. The Kinect is a great example of a disruptive permuation of an old technology because it is fostering an explosion of experimentation that was previously out-of-reach of many people.

When one strays from the purely technical as is necessary here, some bit of crystal-ball-gazing will be required – in some sense I will be venturing guesses as to the collective state of mind of organizations (like CableLabs) and companies (like, say ESPN, or Comcast).  It is with this particularly in mind that I give the following disclaimer – any time I ascribe anything resembling opinion or any other anthropomorphic qualities (or, for that matter, facts unsubstantiated by references) to a third party company, they represent my own opinions.  They are merely what we in the technology industry would term SWAGs – a less polite way of saying “informed speculation”.

A Trip Down the Rabbit Hole (through the broadcast chain)

To begin, the 3D Masters golf tournament is a good case study to follow a 3D video signal all the way from the camera to the living room, and to see how the resolution and motion information is affected.

The cameras used here were pairs of Sony HDC-P1s and HDC-1500s — with the Pace Fusion rig linking them up stereoscopically.  The contribution was 1080i, two HD-SDI feeds per eye.  These were processed in NEP’s SuperShooter (which can originate either 720p or 1080i — pick your poison).  What isn’t clear from media reports, but that we can sort out pretty plainly, is that although the production was financed via ESPN (via Disney, and ultimately via Sony), the broadcast probably did not wend its way to ESPN’s ground station and master control in Bristol, CT, as Bristol has historically been a 720p facility.  At least if it did, it did not use the usual infrastructure.  In turn, its feeds to the rest of the World have, and continue to be, 720p60.  Why they built their programming and facilities around this format back in the day instead of the more typical 1080i is a whole other subject. 

Another wrinkle

A curiosity about this broadcast is that it is one of the few that the cable/MSOs such as Comcast and Cablevision had an exclusive on.  If transmitting 720p60 TaB was the goal of CableLabs, you could be sure that if Comcast had an exclusive on a first-time event such as this, they could have leaned on everyone to tow CableLabs’s line of reasoning.  They didn’t.  The broadcast landed in peoples homes as 1080i, side-by-side.  Or, another way of looking at it, their line of reasoning did…

So, it would seem we either have a 720p-only broadcaster who is very uncharacteristically originating content at 1080i for the first time, OR, we have the largest MSO, Comcast, immediately creating a scism with its own “brother-from-another-mother”, CableLabs.  

My belief is that the answer is mostly the first case, under duress.  The entire broadcast was produced, originated, contributed, and distributed in 1080i.  Just the way I suspect CableLabs and Comcast wanted it, to the opposite purposes of ESPN and the other 720p sports broadcasters.

Why

There is plenty of consternation in the broader industry around CableLabs’ insistence upon TaB for 720p60.  It’s easy to accuse them of ignoring some issues, but the fact of the matter is that these are very intelligent people, and this decision is very likely not without reasons. Here are some possible reasons I came up with – and where the speculation begins:

Drive a stake into the heart of 720p via Sports Broadcast

CableLabs doesn’t particularly like 720p60, because every cable operator dislikes 720p60 (and CableLabs was founded by, is funded by, and exists for the purpose of benefitting MSOs). It’s a complication for settop boxes and other content production, contribution, and distribution in an otherwise harmonious world dominated by 1080i. There is an opportunity to drive a stake in the heart of this format, and CableLabs seems to have taken it.  ESPN has been “irksome”; primarily because of them, all MSOs have to support the entire universe of various ATSC formats most notably 720p.  What better way to “encourage” the House of Mouse and force them to 1080P (or 1080i) than by a Hobson’s choice?

Considering this more carefully, with a twist: 720p60 TaB Stereo 3D exposes the lack of resolution in this format to the point that even the most jaded sports fan is going to notice the lousy resolution (and therefore detail) in a 3D HD broadcast, while those using the more blessed 1080 SbS will only barely notice Comcast Media Center ”gently stepping on” their otherwise HD quality with statmux and allegedly, requant.  TNT Sports shoots and contributes all their sportscasts in 1080i, and this difference will show in 3D like no other.  ESPN will be forced to play ball with Cablelabs, their content will simply be too awful to watch otherwise.  If you are a Programmer, and you want your 3D programs to have carriage on Comcast (or any other Head-End-In-the-SKY — HITS customer) AND you want to contribute 720p?  It has to be 720p60 TaB.  It’s a classic Hobson’s choice.  You could, but the quality will be so utterly hideous and revolting as to thoroughly eliminate it from consideration. Your only REAL choice is 1080i or 1080p if 720p60 SbS is not available as an option for contribution.

For Good Measure, a Silver Bullet for 720p via the 3DTV Market

Just in case the stake in its heart wasn’t enough, here’s a silver bullet.  Imagine a world where instead of shutter-glasses 3D with the attendant $200 per pair costs, a company like Vizio starts mass-producing polarized-glasses displays – wait, that happened at CES in January!  You can bet Comcast and Cablelabs did — these are people who touch the customer repeatedly, and intimately.  They have known for a long time that shutter glasses just will Not Work for a vast majority of households.  Period.  In spite of all the wishful thinking and considerable investment of Panasonic to the contrary.

Polarized glasses can be had for free by watching a RealD theatrical release.  Bought, they cost under a dollar, are disposable, and lend themselves to the “superbowl party at your house” scenario.  Thanks to compnaies like Oakley et. al. you can buy multi-use glasses that serve as both fashion sunglasses AND 3D polarized eyewear.  This will likely be the norm for the mass market, going forward.  Shutter-glasses displays will fail to produce significant mass-market adoption due to major ergonomic issues including flicker when you look away from the 3DTV, weight, dorkiness, and no chance of dual or multi-use.  Not to mention they’re all incompatible with each other.

All of the first generation 3DTV displays require active shutter glasses.  These are electronic devices in and of themselves, require precise engineering, and are very unforgiving to (frequent) design mistakes, and therefore expensive – the Panasonic plasma 3DTV displays, in concert with their glasses, show great 3D… if you can glue them to your face long enough to see 10 minutes of a movie.  The Panasonic glasses are frustrating – they are the best engineered glasses for a 3D experience, but they are very ungainly from an ergonomic point of view.  You simply can’t keep them on your face for the duration of a 90 minute movie, let alone a 4 hour baseball double header.

So how does this affect the SbS and TaB debate?

Given that passive displays will displace shutterglasses displays in very short order (my personal prediction), the Hobson’s choice if not chosen by the programmer — a suicidal choice — will be forced upon the consumer.  Most passive glasses displays divide the left and right eye views into interlaced lines as for “xPol” or line-polarized displays (with the very notable exception of Samsung, who are using RealD’s “zplate” for a full-res polarized solution — the right way to approach this, ultimately).

This means no matter the input signal format, say 1080P SbS, an xPol display will take HALF of the vertical resolution of the left eye, put it into the odd lines of the final display, and half the vertical resolution of the right eye and put it into the even lines of the final display.

Now imagine that a certain MSO’s settop boxes (and a goodly number of consumer electronics manufactuerer’s TV’s) use an xPol or other line-interlaced display technology.  You can imagine that in many circumstances, a 1280×720 video signal, carved up to 720p0 TaB becomes 1280×360 for each eye, then the display takes it and stomps it down to an effective 1280×180 resolution.

To see what this really means, lets take a small subrectangle of a 1280×720 image:

Now let’s fairly decimate it the way a viewer would see it in one eye, with a polarized display, according to 720p60 TaB transmission – assuming a good video processor in the TV:

Finally, let’s fairly decimate it the way a viewer would see it in one eye, with a polarized display, according to 720p60 TaB transmission – assuming a lousy video processor in the TV – and unfortunately, this is the norm (note the loss of resolution, plus the stair-step pixelation):

This is clearly both horrid and repulsive, and all but guarantees that the sports video market will abandon 720p if forced into 720p60 TaB 3D contribution by the MSOs and other distributors who will likely follow the lead of CableLabs.  The first insult would be the forcing of contribution via 720p60 which ensures 360-line effective resolution, even less than that of SD, and approaching VCR.  Second is the insult of the compression to jam 100 pounds of stuff into a 12-pound sack that the Comcasts, Verizons, and DirecTVs of the World use to deliver the video at allegedly-HD resolution, made even worse by 3D.  Finally, the third and final insult would be the likely, market-dominant display technology itself taking an already decimated vertical resolution and decimating it even further to barely-even-YouTube-resolution. 

… And VoD Buries the Evidence

Now, let’s take a different perspective.  Advertising inserts on ESPN and the like, minus the considerable carriage license fees are one thing.  ESPN is appointment-TV, however, and the prime demographic wants to be served what it wants, when it wants it, on the TFT panel it wants it on, be it TV, tablet, computer monitor, and to a lesser extent (this far, in North America) on small mobile, e.g. cellphone.  For my part, I’ve actually watched 3 of my 5 kids crowd around an iPod Nano to watch a television episode. (Seriously, an iPod NANO).  Pay-per-view, and on-demand across multiple platforms is the order of the day, and will be the order of the future. Movies (PPV) and Episodic TV (On-Demand, and Time-shifted via DVR) on any platform anywhere, anytime will be the way younger people consume this content in the present and future.

Let’s examine what this means for HD, and 3D formats in particular.

First, most movie features, whether intended for 3D theatrical, 2D theatrical, direct-to-DVD or direct-to-BluRay, are shot 24P with either digital cameras or actual film cameras.  Many popular episodic TV dramas are shot 24P this way as well (old habits die hard).

Then some variation of what is usually an awful process converts this 24P footage to a 59.94 interlaced 1080 format suitable for broadcast.  The details of this are too gory to go into among mixed company, but suffice to say with few exceptions, it takes smooth motion and makes it various degrees of ugly. Insult upon injury, doing this conversion is not cheap (or, at least doing it well, is not cheap).  In any case, converting this stuff requires labor and capital input something over zero, so it is a prime tendency therefore to scan films only ONCE. Why do it twice if it’s already digitized?

In the meantime, we have to consider 3D — is SbS or TaB better?  Since the original master is progressive, 24fps,  one would think that the simple analysis in the first couple posts (and this one) would apply.  Not so fast.  If you have thousands of VoD programs that have already been digitized and made broadcast-ready via conversion to 1080i, why would you upset the apple cart?  The result is that most films scanned and converted (badly) to 1080i (or 480i) will likely remain that way.  Conversion from interlaced formats to TaB is bad, as covered in the first installment.

For distributors and MSOs, it’s better to stack the deck in favor of content that’s already been mangled to fit the 1080i broadcast chain.  By mandating 720p60 be contributed as TaB, this not only kills the format for broadcast 3D, but utterly ensures its extinction by making Films converted to 1080i and 480i via the industry standard processes for decades, utterly unusable if the Programmer wants to use 720p60.  A Programmer could re-scan the film assets to 720p60, but why? 

Conclusion

In summary — my opinion is that the 720p60 TaB choice by CableLabs is simply this: a solution that is not optimal for the viewer, but is optimal for the MSO along several dimensions — 1) control and leverage over programmers generally so they can control their own intrinsic contribution, distribution, and customer-premise-equipment costs, 2) another lever against Hollywood and other originators, especially Disney and her broadcast properties; and finally 3) a way to give their own VoD content a leg-up cost wise against other libraries that have not been digitized and otherwise would be restrictive economically to re-scan or reprocess.  Particularly clever is that they are using several Consumer Electronic companies as accomplices (whether witting or unwitting).

Nothing is quite as cut and dried as it might seem at first blush.

http://www.cablelabs.com/specifications/OC-SP-CEP3.0-I01-100827.pdf

We covered the intricacies and pitfalls of using TaB (top-and-bottom) format for interlaced sources such as 1080i. As previously mentioned Cablelabs has a couple of figures that visually explain these two alternatives:

Why not keep it simple and use Side-by-Side for all formats then?

To recap, Side-by-Side (SbS) with 720p60 becomes 640×720@60hz per eye. Top-and-bottom (TaB) becomes 720×360@60hz per eye. Both of these are almost standard definition TV resolution. In some cases, the resolution looks closer to VHS than DVD.

So why does CableLabs issue the verdict that ALL 720p 3D CATV should be transmitted TaB? At least 640×720 (SbS) is a nice anamorphic match to the resolution tradeoff of CinemaScope.

As if to add to the confusion, Comcast and several other MSOs broadcast the Masters golf torunament in 1080i SbS format, which was produced by ESPN — who shot the earlier test, and the event itself in 720p60.

How To Uncross Our Eyes Over This

All of these contradictions regarding 720p60 3D transmission seem extremely confusing at first glance — to sort it out one has to peel off all of the layers of the onion, one by one.

The first clue is that CableLabs’ document does not cover the format used in shooting, for initial backhaul to the Comcast Media Center (or other earth station for distribution), nor other intermediate transmissions — just for the CATV system itself, which is to say from the headend to the home.

The second clue is that all of these formats can be converted from one to another; with varying degrees of success and loss of resolution and temporal motion from scenario to scenario.  We saw a pernicious example of this in the previous installment of this series.

The third clue is that almost every North American sports broadcaster: ESPN, Fox Sports, to name a couple, settled on 720p60 as their primary (only) HDTV standard long ago, before 1080p60 began to proliferate. The reasons why this was done are beyond the scope of discussion in this post, but with the exception of TNT Sports, all sports programming in North America is originated in 720p60.

Fourth, in many venues, for many events, we also have to contend with the fact that the camera, switching, tape and replay, and other portions of a program prior to broadcast or uplink termination may also be shot in a different format. Example: a baseball game may be shot with 1080i, shuttled around the broadcast compound, before being converted to 720p60 for uplink.

Fifth, some content does not originate from live events but rather is cinematic content. North American cinematic content is shot at 24fps — sometimes, the legacy content had already been stepped up to 59.94i for broadcast with older analog equipment, sometimes film-to-video transfers of content are available at the original 24fps.

Finally, how the content is actually separated out to YOUR eyes as a viewer by your 3DTV impacts this a great deal. With the latest 3DTV displays, the signal for both eyes is separate, but the actual hardware 3DTV combines them. Various 3DTV displays separate the left and right eye differently, with different mechanisms, advanced engineering, and results. Some use shutter glasses, others use passive glasses (like you get in RealD theaters, like Gucci is producing now)…

Transmission is only half the problem

And this is the key.  To really sort all of this out, we have to analyze the 3D TV display market… and only if we make a very surprising conclusion, then Cablelabs’ push for 720p60 TaB begins to make sense.

This surprising conclusion is for the next installment, when we trace a 720P Stereo3D broadcast from origination to a viewer’s passive-glasses xPol 3DTV set.

It’s pretty obvious that these approaches are two ways to pack two left-right stereoscopic views into a single video frame. Both have the distinct advantage of utilizing the existing HD video transmission and distribution equipment and infrastructure. It’s also obvious that some of the original resolution in the Left and Right frames is sacrificed.There are some interesting implications in their specification, however:

The thinking going into this has several dimensions. These reasons have to do with the overall tradeoffs between horizontal, vertical, and in some cases, temporal resolution of the video formats being used.

A pattern emerges here — why are interlaced signal formats restricted to Side-by-Side? …and progressive formats restricted to Top-and-Bottom? This question and derivatives of it seem to come up frequently, and I’ve seen a variety of answers that seem to miss the real issues.

In the case of progressive video formats, the tradeoff is fairly simple to visualize and explain. When applying one of these 3D formats, you are choosing between sacrificing one half of the vertical resoluton (TaB) or one half of the horizontal resoluton (SbS) in order to squeeze twice the information (a left and right eye image) into a single frame.

Interlaced and Top-and-Bottom do not get along well

With interlaced formats, it’s a more complex question. First, remember that interlaced formats alternatively transmit the even lines, then the odd lines of every video frame, in an alternating sequence. But these sets of even lines, and odd lines — referred to as fields — do NOT join up to form a single frame at a single instant of time the way a progressive frame does. Each field is actually captured by a camera at twice the sampling rate — so moving objects actually register differences with every field. While this means that motion is sampled at twice the progressive rate (e.g. 60 hz instead of 30 hz), each field has half of the vertical resolution.

The loss of vertical resolution is “shared” between odd lines and even lines at alternate times — while doubling the apparent frame rate — the temporal resolution. In interlaced displays, the consequence can be smoother perceived motion, especially for sports and other fast-moving action.

With interlaced video formats, the choice of TaB formatting can have a catastrophic effect. Recall that TaB formatting will sacrifice one-half of the vertical resolution. With interlaced video, one has two choices, none particularly attractive: 1) sacrifice MORE vertical resolution 2) sacrifice temporal resolution.

In the first case, the encoding system takes each field (which is already at half vertical resolution), and throws away every other line (again!).

In the second case, the encoding system throws away every other field and packs it into one or the other of the top or bottom views.

Neither of the two possible ways to carve up 1080i into TaB are attractive — the result is that 1920x1080i at 59.94 fields per second with TaB becomes 1920×260 @ 59.94 effective resolution per eye in the first case. This is an abysmally low amount of vertical resolution. TaB becomes 1920×540 @ 29.97 effective resolution per eye in the second case, which is throwing out half of the motion information. On the other hand, with SbS, 1920x1080i at 59.94 fields per second becomes 960×540 @ 59.94 effective resolution per eye. One does not have to sacrifice temporal resolution (motion is preserved), and is a good tradeoff of spatial resolution.

Why not keep it simple and use Side-by-Side for all formats then?

Let’s consider the 720p60 format. This is 1280×720 with a 60hz frame refresh rate. With SbS, the effective resolution becomes 640×720 @ 60 per eye. This horizontal resolution is actually closer to VHS than DVD, let alone HD, and becomes quite noticeable.

With TaB, 720p60 becomes 1280×360 @ 60 hz per eye. This vertical resolution puts this square into the precise definition of VHS again, not HD… so why do this at all?

This would seem to be especially perplexing for 720P — why not just use SbS for 720P and TaB for 1080P?

Even more perplexing: in spite of CableLabs’ document, Comcast and several other MSOs broadcast the Masters golf torunament in 1080i SbS format, which was produced by ESPN — who shot the event itself in 720p60.

In my next post, I will cover the SbS format in detail – this is also not as simple as it would seem at first blush.

It may not be well-known, but I had the privilege of working with Fox Sports, GameCreek Video, NEP, and Pace, along with the rest of the HDlogix Live Broadcast Team on the Major League Baseball All-Star game in July.  Immediately prior to that game, we were working with much of the same team doing the Yankees/Mariners 3D broadcast from Seattle via DirecTV.

These were, as far as I know, the first examples of true hybrid 2D-to-3D and two-camera-3D broadcasts – where the two were continuously mixed.  Pace provided two-camera stereo 3D capture and production via their Fusion camera systems, and we provided 2D-to-3D conversion of specific camera positions, in addition to format conversion and color correction for a great deal of the graphics, promotional and advertising content which was not in 3D via our ImageIQ systems.

It’s interesting to note that although the Pace crew seemed understandably dubious and wary of our team at first, things thawed considerably when it became obvious that we were both there for the same goal — the technologies serve the content and viewer, not the other way around.  We weren’t there to try any publicity stunts, or to draw compare-and-contrasts between our solutions.  We were there to assist YES network, and Fox Sports in putting together successful shows.

That no one knew which shots were which, indicates to me that not only does the hybrid production model work, but that two production philosophies that have been made out to be at odds with each other can not only work together, but fruitfully complement each other — if planned well — and as long as the quality is high with respect to both sides of the hybrid model.

Both Mike Davies from Fox Sports and Ed Delaney from YES Network had good things to say about it.

HDlogix is a relatively new startup, just 2 years old. Our company is really the result of several investments and acquisitions going back some 10 years, however. So, it’s likely that you have never have heard of us, but it is probable that the pixels you’re now staring at through your LCD monitor are running through our patented — and licensed — processes and hardware.  Additionally, we have the benefits of a deep and applied IP portfolio and seasoned R&D team that have been working together on advanced video technologies for well over 5 years, on average.  In short, HDlogix could probably not be characterized as a typical startup.

We realized that some of the very complex technology we had developed could be applied to a 2D-to-3D video process, and with the intense interest of some of our largest customers pushing us forward, we decided to shift the company focus from video processing, scaling, and format conversion to 3DTV and 3D cinema.  Additionally, we had begun to focus on advantaging the astonishing amount of processing power in GPU video chipsets — like those from ATI or NVidia — and use those to expand the complexity, and therefore the quality of our video processes.  Designing and implementing processes that can run on hundreds, or even thousands of processing cores has been our bread-and-butter for many years; meanwhile, many others are still struggling to figure out how to take advantage of 8 cores on today’s PC systems.  The sorts of techniques that others wouldn’t even think of attempting, because  they would be so painfully slow at SD video resolutions as to be unusable, we can run — in real-time, at DCI resolutions, such as 2K (2048×1080 and 2048×1556 anamorphic), or near real-time at 4K (4096×2160).  ATSC 1080i is obviously not a problem…

There are a lot of deserved reservations about “automatic 2D-3D conversion”.  I say “deserved” because many of the existing processes may work well for some types of content, or some formats, but have other limitations that make them wholly unsuitable as a high-quality and robust solution in a broadcast or professional environment.  The 3D industry has seen well more than its fair share of snake oil, and as a result, savvy industry participants are very skeptical of grand claims. 

The primary difference between HDlogix and all others is that other solutions have been engineered to fit within a few thousand transistors, or very few CPU core pipelines.  Our solution literally uses billions of transistors, and teraflops of parallel computing power for a single video stream to achieve its results.  At the end of it all, what matters is the perceptual quality of the depth cues, the prevention of any eye fatigue, and ensuring that the overall quality and resolution of the video does not suffer.  The ability to achieve these goals are what makes HDlogix’s ImageIQ3D solution stand alone.

ImageIQ3D also corrects native 3D video and film content to reduce viewer fatigue and discomfort by fixing the problems introduced by stereo camera limitations, operator oversights, editing, shot-selection mistakes, and post-production processes.

You can read about the ImageIQ3D process in excrutiating detail here: http://hdlogix.com/modules/text-big/collateral/HDLogix_3D_IQ_Overview001.pdf

Stay tuned…

Apparently, even your bathsoap is now in 3D:

Seriously? 3D SOAP?

Now, I’ve seen things like the “Banzai 3D Shark Bite Water Slide” which is essentially a pretty ordinary kids water slide, except it has cyan and red anaglyph figures printed on it, and includes a single pair of red/blue anaglyph swim goggles — from my own kids’ reaction to the commercial early this summer, they just had to have it.  I wasn’t terribly impressed, but my kids were.

Kids! Get Your Cheap 3D Thrills Here!

What’s your opinion? Is the 3D Hype getting to be too much?