LR format

	Re: LR format by Chris Eller on Sep 27, 2009 at 4:42:39 pm Until SMPTE actually sets the stereo standard, you're going to get answers and recommendations all over the road. For example; I can freeview both parallel and cross-eyed, and I prefer to format side-by-side content (when needed) as L|R for logical reasons. I think the left video should be on the left side of the screen, it's easy to remember when this is the case. Playback devices, for the most part, have the ability to "swap eyes" so you can actually encode however you wish and if there is a problem swap eyes. This is not ideal of course, we should be able to encode to a standard and it should just work across any type of viewing technology. Until SMPTE actually sets the spec, we're left in this gray area. Seconding an earlier post, when possible encode the full frames and don't mess with anamorphic squeezing, you're already losing resolution on most displays (checkerboard DLP and Xpol being the most prevalent). Use the WMV video layers capability and free utility from Peter Wimmer at http://www.3dtv.at/. This should reasonably future proof you and retain the highest quality. The 3D BluRay proposed spec is double frame rate interleaved: L|R|L|R|L|R|L|R|L|R|L|R. For some reason the YouTube engineer working on the YouTube 3D player chose to have the native layout as R|L and we're stuck with this de-facto standard for the time being. I encode L|R and use the yt3d:swap=true flag for the time being. By the way, how can we post a YouTube link without the embedded player function? The embedded player does now show the YT3D viewing options. * * If anyone from SMPTE is reading these fora, please set the spec sooner rather than later. Thanks! -- Chris Eller * Advanced Visualization Lab - Indiana University * Chris Eller Photography, LLC * Starrynight Productions, Inc. Respond to this post   •   Return to posts index

	Standardizing on standards, Re: LR format by Tim Wilson on Sep 28, 2009 at 2:17:38 pm [Chris Eller] "If anyone from SMPTE is reading these fora, please set the spec sooner rather than later. Thanks!" We have hundreds of SMPTE members who regularly visit these forums, including quite a few regular posters. This deserves an entire thread, but the thing about standards is...there's often more than one. Look at HD - about half the networks produce 720, and half 1080. The standards primarily speak to interoperability....which is critical...but it's not as if only one of them is usable, or even optimal. In fact, I'd argue that 720 is more often more useful for now. The point being that SMPTE has created a SET of standards for HD. Absolutely not just one. Note too that SMPTE only covers broadcast, and does zero for standards related to digital cinema (most typically displayed at 1920 or less, making HD standards very much an issue), and had zero to say about Blu-ray or HD DVD. Which points to the ways in which more than one GROUP of standards is coming. SMPTE will only speak to the mastering format - the working group has started with 1080/60 per eye - but I don't get the impression that it will necessarily pick one and only one way to capture or encode it -- only the mastering itself. Here's the thing: most displays simply don't care. HDMI will simply dump out a pile of pixels, and the displays will show it. As long as they know what to look for, the chips feeding the HDMI output can also handle whatever is thrown at them - just as they know how to handle mixtures of 720, 1080, quarter frame or half frame hardware upconversion, progressive, interlaced...oh yeah, and SD. And none of this says that you should be using any of several flavors of HDV, DVCPRO HD, XDCAM, HDCAM, or even that you can't slip some 1440 in there....and quite often, some combination of all of these in a single project. Oops, and different color encoding and frame rates for other parts of the world. Again, SMPTE only helped narrow groups of choices that are still pretty darn big. In the Thinking Big issue of Creative COW Magazine (coming Oct 09), we'll be talking to Ethan Schur of TDVision. They've got a codec that handles 1080p/60 using a delta compression (mathematically lossless by capturing only difference pixels in the second eye), and can natively handle side by side, top-bottom, interleaved, checkerboard, and even anaglyph: tell HDMI where the pixels go, and that's it. Theirs is one proposal - it happens to be the one I find most compelling - but the point is that it allows any and all methods of current standards of 3D encoding...just like current decoders can handle all the HD and SD standards I mentioned. If SMPTE is as excited by this as I am, it will allow producers to choose their own methods for production and post, and codecs will handle the rest for distribution and display. This, frankly, is exactly how it SHOULD work. So I can easily imagine a set of standards that excludes none of these, as manufacturers press to make sure that they are included. I can also imagine an evolution, as with 720 to 1080 - which, ironically, supported interlaced BEFORE progressive, which is lower bandwidth. However many years along we are, you can buy displays with all of them. Here are some of the other bodies who will be working on different parts of the pipeline: 3D Home Master uncompressed format: SMPTE 3D Home Distribution (encoding/decoding) format: 3D@Home, Advanced Television Systems Committee (ATSC), Society of Cable Telecommunications Engineers (SCTE) 3D Encoding format: International Telecommunication Union (ITU), International Electrotechnical Commission (IEC), International Organization for Standardization (ISO), and the Moving Picture Experts Group (MPEG) 3D Interfaces and visualization devices: Consumer Electronics Association (CEA) The interfaces one really struck me. There will need to be a way to standardize the display of onscreen information - open and closed caption, viewer choices, menus, etc. - in ways that are meaningful in 3D space, or else you won't be able to easily read them! You might think that simply laying them on top of the 3D information, but, thanks to floating windows (which TDVision's codec already supports), the notion of "on top" changes - onscreen controls that don't also support that could break the underlying 3D. Or create a situation where the controls appear to also be moving forward and back in space. Or really truly floating out front while their relative distance from the background keeps changing. It would create the illusion that the size of the controls are changing! This is obviously not a problem for shooting, which this forum has so far primarily focused on. But it might sometimes be a problem for post folks who have to incorporate these kinds of design elements, and a WHALE of a problem for set-top box developers. Anyway, there will be half a dozen or more "standards" beyond SMPTE, virtually none of which will including only a single vector, that YOU will have the pleasure of managing. So what's the answer for now? It will be the same answer after some of these standards are defined: do whatever the client pays you to do, and let them worry about what to do with it. :-) Yr pal, Timmy Tim Wilson Creative Cow Magazine! My Blog: "Is this thing on? Oh it's on!" Don't forget to rate your favorite posts! Respond to this post   •   Return to posts index

	Re: Standardizing on standards, LR format by Chris Eller on Oct 12, 2009 at 2:26:31 am Tim, Thank you so much for this post! It really helps to have a good idea of what is going on behind the scenes. I had no idea that there were so many groups involved. I've heard of all the groups, but did not know how their work interleaves. So, for what it is worth, here's my packaging plan for the time being: * Sony EX3 Rig ** record 1080p per camera (24 fps the current favorite) ** edit and output two complete WMV files ** Use Peter Wimmer's wmvmux.exe utility to layer the videos into a dual layer .wmv file *** Left first, right second * Fujifilm FinePix 3D W1 camera (video mode) ** record 640x480 30p per eye ** automatically saved as a layered AVI file *** I've not tried editing the resulting video file yet, have to do that soon. Results: * plays left eye in mono video applications at full resolution * plays stereo in stereo video apps at full resolution per eye ** The display may downsample this; 3D-DLP, Xpol Again, thanks for the insight Tim. -- Chris Eller * Advanced Visualization Lab - Indiana University * Chris Eller Photography, LLC * Starrynight Productions, Inc. Respond to this post   •   Return to posts index