Passive 3D; Which is better: T/B or SbS?

[fritzi93]

Wasn't sure which forum to post this in. Anyhow...

I've read some discussions over at the AVS forums about which is better for passive 3D: top/bottom or side-by-side.
One line of reasoning is that T/B is better than SbS because you're looking at half-height resolution anyway.

That makes no sense to me. Can some one please explain? Thanks.

[pandy]

http://www.sisveltechnology.com/files/3Dresolutionevaluation.pdf

http://www.cablelabs.com/specifications/OC-SP-CEP3.0-I02-110131.pdf (page 23)

Additional explanation - TaB is allowed only for progressive video.
(chroma sampling scheme for 4:2:0)

[fritzi93]

Thanks, those pdf files are helpful. I slogged through them once, and will have to read them several times.

I do SbS captures from cable box with my HD-PVR. SbS is exclusively 1080i, and obviously that's all you'll get with cable at that resolution. No TaB content is available from my cable provider, which would have to be 720p anyway. TaB 1080p will be a do it myself deal, from 3D BDs. Just out of curiosity, and to play around a bit with encodes and see how they look on my TV.

Interesting point in the first document as to the eye's sensitivity to horizontal versus vertical resolution. No consensus there.

If you leave aside whether image fusion works or not, it seems to me passive 3D does have a couple drawbacks: The black matrix mask is a little thicker (and more visible if you get very close to the screen), and the TV is prone to moire and jaggies when displaying sharp horizontal lines. On the other side of the ledger, practically no ghosting, flicker, crosstalk (except at vertical offset), wide viewing angle, bright picture.

It just seemed to me that someone saying TaB is better on a passive 3D TV because of supposed halving of vertical resolution *inherent* in said TV is a non-sequitur. What am I missing here?

Thanks in advance for any more replies.

[pandy]

btw can You explain using term passive 3D (passive in which way - coding or displaying?)

[fritzi93]

Passive display, aka FPR (film pattern retarder). LG calls it Cinema 3D, and it uses glasses identical to RealD circularly polarized glasses you get in the theater.

[Cornucopia]

Let me see if I can shed some extra info (and my own spin) on this...

The 1st document only discusses "Frame Compatible" (aka existing 2D-compatible) options, excluding Anaglyph. This disregards the up-and-coming change to ATSC, DVB and other corporate technology strategies that will allow "Service Compatible" options, such as MVC encoding or Dual-stream encoding (aka Simulcasting) using MPEG4 AVC/MVC or even h.265 formats instead of MPEG2, and piping it to the TV with "Frame Packing" SEI. This will make many of the previously narrow options of non-Full3DHD moot.

The second document pre-supposes that the broadcaster or other corporate source is using only "OpenCable" systems standards and unfortunately it also (incorrectly) presumes that Side-by-side is incompatible with Progressive encoding (the reverse is mainly true: Top/Bottom is usually incompatible with Interlaced encoding). This is VERY restrictive, but since this is a standards body defining their standard, if people expect to follow it, those will be the available options (narrower as they might be).

However, fritzi93, using HTPCs or certain (particularly Chinese-made) media players, these OpenCable standards are not adhered to, so you have more options than that. But also greater compatibility uncertainty.

But to answer your original question:
Particularly, if you are using original material (created by your own camera, for instance), it is MUCH better to MATCH the resolution strengths of the Source Format, with the Storage Format and the Display Format.
If you have compromised resolution in one dimension, you want to keep that compromise, not compromise the image FURTHER by halving it in the other dimension.
So if your source is Top/Bottom, your greatest resolution is in the Horizontal, and you don't want to cut in half the Horizontal resolution by storing it as Side-by-Side.

However, Passive displays are mistakenly considered like Interlaced displays - THEY AREN'T! Both images exist on the display simultaneously, not one after the other (people are confusing this with Active Displays, which ALSO aren't line-interlaced but Full Frame sequential).
Since Passive displays show the full 2 images, yet each full image is Full Horizontal and Half Vertical resolution, this is ideally suited to Top/Bottom images. It's just a simple re-interleave operation.

Therefore, Top & Bottom could be considered the most "natural" format for Passive Displays.

Now, what's interesting that most 3D experts seem to keep ignoring is the fact that Stereo 3D's Depth dimension (its "Z" plane) is not directly defined like it is in 3D animation or real-world techniques, yet it does exist. It is calculated and understood to be a direct correlation between the FINE-NESS in the binocular disparity (parallax) between the Left & Right views. And binocular disparity is only calculated on the Horizontal dimension!

So that actually means that a format like Top/Bottom, even though it does greatly lose resolution in the Vertical Dimension, actually has better resolving power (twice as fine) in the Z-Dimension than does Side-by-Side. So it really makes sense to use Top/Bottom on a Passive Display whenever possible, in order to get the best 3D benefit.

The problem is that greater than 75% of Frame-compatible 3D distribution (via Broadcast/Sat/Cable, etc) is in Side-by-side format, mainly because HD channels are more often 1080i rather than 720p.
Since you are given Side-by-Side, you are stuck with reduced horizontal, regardless of display. And in those cases, Passive displays are the loser compared to Active Displays, WRT 3D resolution.

Luckily, even 1/2 V rez + 1/2 H rez 3D is still good enough to watch & enjoy! So I wouldn't sweat it too much.

Scott

[fritzi93]

Now that's what I was looking for. A lot to mull over there. Thanks a bunch.

[pandy]

Passive display, aka FPR (film pattern retarder). LG calls it Cinema 3D, and it uses glasses identical to RealD circularly polarized glasses you get in the theater.

Ok then this is exactly the same as i understood - display technology not coding.

Example for other ideas of the frame packing - this one "invented" by Sisvel http://www.sisveltechnology.com/news/files/3Dwhitepaper.pdf

Let me see if I can shed some extra info (and my own spin) on this...

The 1st document only discusses "Frame Compatible" (aka existing 2D-compatible) options, excluding Anaglyph. This disregards the up-and-coming change to ATSC, DVB and other corporate technology strategies that will allow "Service Compatible" options, such as MVC encoding or Dual-stream encoding (aka Simulcasting) using MPEG4 AVC/MVC or even h.265 formats instead of MPEG2, and piping it to the TV with "Frame Packing" SEI. This will make many of the previously narrow options of non-Full3DHD moot.

Or not - real MVC technology (free viewpoint TV) seems to be not realistic in nearest decade, i dont think that also we will see MVC (L + R) with central synthesized PoV.

The second document pre-supposes that the broadcaster or other corporate source is using only "OpenCable" systems standards and unfortunately it also (incorrectly) presumes that Side-by-side is incompatible with Progressive encoding (the reverse is mainly true: Top/Bottom is usually incompatible with Interlaced encoding). This is VERY restrictive, but since this is a standards body defining their standard, if people expect to follow it, those will be the available options (narrower as they might be).

My point was that Cablecom is large company and usually those large companies create some more or less but accepted worldwide standards - of course there are alternatives and we always can "invent" something much "better or not" like for example in Sisvel paper i've provided above.

[Cornucopia]

Passive display, aka FPR (film pattern retarder). LG calls it Cinema 3D, and it uses glasses identical to RealD circularly polarized glasses you get in the theater.

Ok then this is exactly the same as i understood - display technology not coding.

Example for other ideas of the frame packing - this one "invented" by Sisvel http://www.sisveltechnology.com/news/files/3Dwhitepaper.pdf

Let me see if I can shed some extra info (and my own spin) on this...

The 1st document only discusses "Frame Compatible" (aka existing 2D-compatible) options, excluding Anaglyph. This disregards the up-and-coming change to ATSC, DVB and other corporate technology strategies that will allow "Service Compatible" options, such as MVC encoding or Dual-stream encoding (aka Simulcasting) using MPEG4 AVC/MVC or even h.265 formats instead of MPEG2, and piping it to the TV with "Frame Packing" SEI. This will make many of the previously narrow options of non-Full3DHD moot.

Or not - real MVC technology (free viewpoint TV) seems to be not realistic in nearest decade, i dont think that also we will see MVC (L + R) with central synthesized PoV.

The second document pre-supposes that the broadcaster or other corporate source is using only "OpenCable" systems standards and unfortunately it also (incorrectly) presumes that Side-by-side is incompatible with Progressive encoding (the reverse is mainly true: Top/Bottom is usually incompatible with Interlaced encoding). This is VERY restrictive, but since this is a standards body defining their standard, if people expect to follow it, those will be the available options (narrower as they might be).

My point was that Cablecom is large company and usually those large companies create some more or less but accepted worldwide standards - of course there are alternatives and we always can "invent" something much "better or not" like for example in Sisvel paper i've provided above.

@pandy, that "Frame Packing" is totally different from HDMI's "Frame Packing" format. The former is incorrectly named "service compatible" when it actually is "frame compatible" in the same way that current T/B or SbS is. All it requires is a slight firmware update for the existing MPEG2/MPEG4/h.264 decoders in the tuner/receiver boxes (or other settops) to understand the difference between an orignal image dimension & its "bounding rectangle" or "allowable window", and then even those 2d devices would be immune to misunderstanding the signal and mis-displaying it in its original (mixed) layout (which is better than can be said for current T/B or SbS).

The Latter format is NOT Frame Compatible as the signal is (usually) 1920x2205, which is like a FullHD x 2view Top/Bottom 1920x1080p, but with 45 pixel height buffer between images. This is obviously BIGGER than is currently acceptable with ATSC, DVB, BD, etc. It is generated on the fly by BD3D players after they've decoded & composited the MVC signal.

BTW, the Former Frame Packing format you mentioned is of limited used, since it requires a 1080 channel, and since most are Interlaced, what you'd get in the output would be only 2views x 720i - are really awkward (orphaned?) format.

***************

Also, I was not referring at all to Free viewpoint coding, just the standard everyday MVC/SVC that uses Main(base)+Differential(dependent) views (2 views), which you can find in every BD3D and on a number of semipro 3D cameras. I'm fully aware of the extended gestation period of free viewpoint, just as I'm understanding of the long-term strategies for integral imaging and Realtime projection holography for use as 3D vehicles.

***************

I agree about the OpenCable spec, but even now it's not even close to being universally adopted, and there's always the chance for them to come out with a version 2 which is less restrictive. Also, the assumption that 4:2:0 has any effect on limiting use of T/B is false. A frame compatible 1920x1080 is 1920x540 per view. 540 is still Mod4. Same with 1280x720 being 2view x 1280x360 still Mod4. That means that both image types have frame boundaries that DO NOT interact with each other WRT 4:2:0 sampling, even in Interlaced mode. That's a red herring.

Scott

[pandy]

Oh - my point was that : "paper is very patient - it will accept anything, even most stupidest ideas" - it is only to market what "standard{" is accepted and market is usually created by large companies, way how You encode, way how You organize display view can be very flexible but at some moment You can have "best technology" of the World and at the end You are ignored because market has own idea how to solve problem (not always in cheapest or most elegant from engineer point of view way).

And , yes, i agree with You, but nowadays quality for broadcast lowered to such point where large, public operator, broadcasting their services with incorrect field dominance (this is common in Europe) - public broadcasters - for decades offers highest quality standards and today they don't care...