MTBS-TV: Interview with Richard Laberge, Sensio Co-Founder

[Neil]

[youtube-hd]http://www.youtube.com/watch?v=tCK4_WL9URc[/youtube-hd]

Richard gives a primer on what Sensio does, explains how their technology is implemented in stereoscopic 3D gaming, and outlines their latest product lines and partnerships. Very interesting! Share your thoughts.

Regards,
Neil

[yuriythebest]

cool! commented on youtube :polar

[cybereality]

Very nice.

[taz291819]

Really nothing new in that interview, and he didn't explain things correctly. HDMI 1.3 does have the bandwidth for Full 1080p 3D, it just doesn't have the protocol for transporting said video. He mentioned HDMI 1.3s bandwidth issues several times, which simply isn't true.

[DmitryKo]

HDMI 1.3 does have the bandwidth for Full 1080p 3D, it just doesn't have the protocol for transporting said video

If you're referring to posts made on the AVS Forums, these calculations are simply not incorrect - both HDMI 1.3a and 1.4 are limited at 340 MHz pixel clock, which amounts to 8.16 Gbit/s of video bandwidth. My own calculations show that the video bandwidth of 1920x1080 x24 bpp @120 Hz mode would clearly exceed the effective bandwidth of both HDMI 1.3a and dual-link DVI, at least when using standard CVT formula. Materials published on HDMI website offer conflicting information regarding 120 Hz mode support; the freely accessable HDMI 1.3a specification (2006)refers to CEA-861-D (EDID monitor profiles for DTV) for a description of available video modes.

The newest revision of the EIA/CEA-861 standard is CEA-861-E, adopted in 2008.Wikipedia article on EDID says that 1080p120 mode was only added in this 2008 revision - which means HDMI 1.3 devices and cables were never, ever designed and tested to handle 1080p120, and even HDMI 1.4 defines full-resolution frame sequential modes as optional. I don't have acces to the full text of the standaed and cannot really tell what video timings and how much bandwidth do the two 120 Hz modes use. But if you check user manual for acclaimed professional testing instruments like Quantum Data 881/882 http://www.quantumdata.com/products/882E.asp" onclick="window.open(this.href);return false; , HDMI models don't have any 120 Hz formats in their standard format library.

He mentioned HDMI 1.3s bandwidth issues several times, which simply isn't true

Yeah, and all those brands who license Sensio and RealD formats are outright stupid and cannot even listen to their own engineeres, including Texas Instruments who chose the checkerboard frame packing format for their 120 Hz DLP TVs, and even HDMI Founders only support half-resolution frame packing formats as a common denominator in their HDMI 3D spec... would make a good conspiration theory.

[taz291819]

HDMI 1.3 does have the bandwidth for Full 1080p 3D, it just doesn't have the protocol for transporting said video

If you're referring to posts made on the AVS Forums, these calculations are simply not incorrect - both HDMI 1.3a and 1.4 are limited at 340 MHz pixel clock, which amounts to 8.16 Gbit/s of video bandwidth. My own calculations show that the video bandwidth of 1920x1080 x24 bpp @120 Hz mode would clearly exceed the effective bandwidth of both HDMI 1.3a and dual-link DVI, at least when using standard CVT formula. Materials published on HDMI websietoffer conflicting information regarding 120 Hz mode support; the freely accessable HDMI 1.3a specification (2006)refers to CEA-861-D (EDID profiles for DTV) for a description of available video modes.

The newest revision of the EIA/CEA-861 standard is CEA-861-E, adopted in 2008.Wikipedia article on EDID says that 1080p120 mode was only added in this 2008 revision - which means HDMI 1.3 devices and cables were never, ever designed and tested to handle 1080p120, and even HDMI 1.4 defines full-resolution frame sequential modes as optional. I don't have acces to the full text of the standaed and cannot really tell what video timings and how much bandwidth do the two 120 Hz modes use. But if you check user manual for acclaimed professional testing instruments like Quantim Data 881/882 http://www.quantumdata.com/products/882E.asp" onclick="window.open(this.href);return false; , HDMI models don't have any 120 Hz formats in their standard format library.

He mentioned HDMI 1.3s bandwidth issues several times, which simply isn't true

Yeah, and all those brands who license Sensio and RealD formats are outright stupid and cannot even listen to their own engineeres, including Texas Instruments who chose the checkerboard frame packing format for their 120 Hz DLP TVs, and even HDMI Founders only support half-resolution frame packing formats as a common denominator in their HDMI 3D spec... would make a good conspiration theory.

We're talking about two different things. You said it yourself, both HDMI 1.3 and HDMI 1.4 are both limited to 340Mhz. If HDMI 1.3 doesn't have the bandwidth, how can HDMI 1.4 have the bandwidth?

1080p60 (24-bit) uses ~3.56Gbps x 2 is 7.12Gbps for Full 1080p 3D. The maximum bitrate for let's say, Dolby TrueHD is 18 Mbps. This total is well under the bandwidth constraints of HDMI 1.3.

http://nslog.com/2006/11/18/1080p_math" onclick="window.open(this.href);return false;

EDIT:
I read your link, and don't quite understand why vertical blanking is necessary with Blu-Ray. For broadcast, or course, I know exactly what it's used for, as I work in television broadcasting. If you could explain, I'd appreciate it.

And I'm not knocking Sensio's method (or RealD for that matter), as broadcasters will have to "frame pack", there's no way around it, unless they plan on swapping out everyone's STBs.

[DmitryKo]

You said it yourself, both HDMI 1.3 and HDMI 1.4 are both limited to 340Mhz. If HDMI 1.3 doesn't have the bandwidth, how can HDMI 1.4 have the bandwidth?

Does it have the bandwidth? 8) The fact that HDMI 1.4 requires CEA-861-E and the latter defines 1920x1080 120 Hz mode does not mean that HDMI devices can support this mode - every mode is optional, even the "standard" ones, and besides that single 120 Hz mode, there are probably enough other important revisions in the document to warrant its use.

If the Quantim Data 881/882 example was not enough, here is the latest TDMS transmitter chip from Silicon Image that supports HDMI 1.4 3D formats, http://www.siliconimage.com/products/pr ... px?pid=153" onclick="window.open(this.href);return false; I don't see any support for 720p120, CEA-861-E and 1080p120 in the specs.
Another 1.3a compliant device http://www.nxp.com/documents/data_sheet/TDA19978A.pdf" onclick="window.open(this.href);return false; that only supports 235 MHz clock, enough for 36-bit 1080p60.

1080p60 (24-bit) uses ~3.56Gbps x 2 is 7.12Gbps for Full 1080p 3D

You can not multiply it this way, because guard intervals are defined in seconds and so total pixel resolution will more than just 2 times higher. Please check http://vesa.org/Standards/free.htm" onclick="window.open(this.href);return false; for XLS spreadsheets which can be used to calculate the timings by GTF, CVT and CVR (reduced blanking) formulas, which should give you the idea.

Considering that standard CEA-861 60 Hz mode 16 has a total of 2200 x 1125 pixels (something in between CVT-standard 2576 x 1120 and CVR 2080 x 1111 pixels), 1080p120 could in fact fit into 8 Gbit/s. Unfortunately, I can't find the timings online because the license terms for CEA-861-E disallow anyone to publish the text of the standard, and EDID tools for getting the Detailed Timings Data block would be useless because there is no 1080p display capable of taking 120 Hz input.

don't quite understand why vertical blanking is necessary with Blu-Ray

It has nothing to do with the image source, be it Blu-ray or video game, blank lines are generated by the transmitter logic. NTSC/PAL, CEA/EBU/SMPTE (TV, DTV and HDTV), VESA (computer display), military and medical video modes have always been defined with blanking for the last 55 years. What would be a good reason for dropping it and making all current displays go straight to the trash can, except for a minor bandwidth gain?

The practical pplication is tranmitting audio data in blanking intervals of HDMI Data Island periods.

BTW pixel clocks for most common 50 and 59.94/60 Hz DTV/HDTV modes were artificially made equal by adding more blanks to the 50 Hz signal. I can't understand wWhat's up with that requirement.

[DmitryKo]

1080p60 (24-bit) uses ~3.56Gbps x 2 is 7.12Gbps for Full 1080p 3D

You can not multiply it this way, because guard intervals are defined in seconds and so total pixel resolution will more than just 2 times higher

On a second thought, after exploring standard CEA-861 timings for 720 and 1080 signals at i50, p50 and p60 scan rates, as well as some standard VESA timings (courtesy of the above-mentionet datasheet from NXP), it looks like due to the way CEA-861 constructs its video modes it is perfectly possible that 100/120 Hz and side-by-side modes actually can be derived from 50/60 Hz modes, and the bandwidth can be multiplied by 2.

As I understand it, both frame-sequential 1080p100/p120 AND side-by-side 1080p50/p60 modes should each consume 7.13 Gbps in 24-bit pixel mode, perfecky within the advertised limits of HDMI 1.3/1.4 with Category 2 cables (see Common stereoscopic video resolutions and bandwidth for details).


Still, it does not necessarily mean that there would be any products utilizing any of these modes in the near future.

[taz291819]

Still, it does not necessarily mean that there would be any products utilizing any of these modes in the near future.

That we can all agree on, there's no telling what they're going to do.