Why is checkerboard 3d inferior to standard 3d?

[bastian74]

A note on the Samsung DLP TV's -- the picture is slightly defocused, using what they call "wobulation." This tends to blur the checkerboard images enough that they look like a full-res image (at the cost of some maximum sharpness).

I'm not sure this is very effective. If you draw a 45 degree angle line in paint only one eye will be able to see it. Thin objects moving on the screen "shimmer" too.

[chrisjarram]

There is more info here (though a fair bit of BS in this article, whoever wrote it should be shot!!)

http://www.pcmag.com/article2/0,2817,2338401,00.asp" onclick="window.open(this.href);return false;

For me, this is an absolute deal breaker either way - think I'll just stick with my 'superior' xr-10x which cost a fraction of the price.

[pixel67]

hmmm. My old Infocus X3 could do full page flip at 85hz.

[chrisjarram]

hmmm. My old Infocus X3 could do full page flip at 85hz.

Exactly, there are so many projectors out there much cheaper, and for this reason much better, than the new InFocus - they have totally shot themselves in the foot making this a checkerboard unit afaic. I'd MUCH rather have full pageflipped resolution that the extra 35hz - 85hz I can use for unlimited periods of time with no eyestrain whatsoever.

[BlackShark]

So if you draw a horzontal 1pixel wide line on the screen only one eye will be able to see it?

If you send this 1 pixel horizontal line to the display in 3D mode : yes
This is why small text is usually impossible to read on interlaced displays.

[smoothy]

There is more info here (though a fair bit of BS in this article, whoever wrote it should be shot!!)

http://www.pcmag.com/article2/0,2817,2338401,00.asp" onclick="window.open(this.href);return false;

For me, this is an absolute deal breaker either way - think I'll just stick with my 'superior' xr-10x which cost a fraction of the price.

I just read that article, a few things I would like to point out which have left me wondering with questions

It said cut and paste 'Currently the most common approach — used by both IMAX and the ViewSonic projector — is to combine alternating images for the left and right eyes with glasses built around LCD shutters. The glasses synchronize with the projector so each eye sees only the image meant for that eye. The synchronization is maintained either using pulses of light on the screen generated by the projector or (as with the PJD6220-3D) infrared light generated by additional hardware near the screen.'

Does that mean imax uses shutter glasses? I always thought they were using circular polarized glasses? The above cut and paste did mention pulses of light on screen. Could it be imax uses the pulses of light with polarized glasses?

is 1400x1050 a widescreen resolution?

Can this viewsonic projector only do 120hz 3d in 1024x768 and say 3d at 85hz in 1400x1050?

I have heard it cannot do high definition but 1400x1050 is high definition it can correctly display 1280x720p it won't be native but you can have that content with say 2d movies displayed at 1400x1050 right? It can also accept 1080i but that will be down scaled to 1400x1050 or will it go to it's native resolution of 1024x768?

It does checkerboard 3d in it's native resolution of 1024x768 at 120hz, now the big question is can it do full resolution 3d at say 1024x768 or 1400x1050 at 85hz+?

Sorry for all the questions

I hope you guys can clear these things up which would help anyone else that have any questions about this projector

Many many thanks in advance

[BlackShark]

Any resolution can be widescreen providing you use an anamorphic pixel ratio (non-square pixels)

Some 3D theaters use shutter systems, some use circular polarisation, others use dolby3D.
I'm not sure if there is a IMAX norm about this.

3D theaters do not use the checkerboard pattern but some of them use shutter glasses.

[Neo42]

My understanding of checkboarding is this:

Player takes 2 full 1080p frames in 1/30th of a second, one for left eye and one for right eye, and breaks them into 4 new frames, the first one being left eye half-res checkerboard, the second being right eye half-res checkerboard and the following 2 frames are the other half checkerboards for those frames. My assumption is each frame sent to the display is actually 2 half frames 'interlaced' to be split for each eye, with the following frame the other half of the image for each eye. The net effect is: half image left eye, half image right, second half image left, second half image right. This sequence happens 30 times or 120 individual partial/half frames per second. Aside from this, depending on the implementation, the display actually does some processing to interpolate the missing pixels in each single-eye sub-frame to reduce the perception of the checkerboard pattern.

If this explanation is accurate, the end effect is 2 full 1080p frames (one for each eye) in each 1/30 sec slice, a derivation of 1080p60. In any given frame yes there is half the resolution, but overall I don't consider this a 'loss of resolution' at all. So is there really any 'loss' of resolution, practically speaking?

[DmitryKo]

Player takes 2 full 1080p frames in 1/30th of a second, one for left eye and one for right eye, and breaks them into 4 new frames, the first one being left eye half-res checkerboard, the second being right eye half-res checkerboard and the following 2 frames are the other half checkerboards for those frames.

This is not correct. What you describe is a variation of 1080p30 format; however similar 4-field scheme is used in 1080i stereo, except that each subfield is not checkerboard sampled but interlaced (vertical subsampling).

In TI's checkerboard 1080p60 format, the left picture is always sampled at odd positions ("white" cells on the chessboard), and the right picture is always sampled at even positions ("black" cells on the checkerboard). The result is 1080p60 video stream that contains two half-resolution left-right frames in each 1080p frame. In order to display the stereo video on a frame-sequential stereoscopic display, each left-right frame is up-sampled to full-resolution 1080p by averaging the neighbouring pixels (aka bilinear interpolation). The result is 120 Hz video which contains alternating left-right frames, each at 1080p60, presented with the help of LCD shutter glasses.

Side-by-side checkerboard is another variation which is a half-resolution side-by-side format (horisontal subsampling) where each sub-picture can be checkerboard sampled, as explained in the 3D portion of HDMI 1.4 spec. This format is more flexible and left-right pictures can be sampled at different combinations of odd and even sampling.

[Neo42]

DmitryKo-

I think I just got mixed up on the streaming format coming out of the player/display adapter-- each frame contains 2 offset checkerboard half-res images, or a combination of left and right eye images. The following frame will be the other halves 'essentially' giving you 30 full frames per eye, per second or 1 half frame for one eye every 1/120th of a second (alternating between each eye)

[BlackShark]

This isn't the DLP checkerboard indeed, it sounds like the sensio format that we've been able to see in Avatar the game.

[Neo42]

I've played Avatar in 3D on PS3 via DLP checkerboard and found it to look quite good.

I think I would rather have 120hz checkerboard over 60hz full frame as the latter would probably give me a tremendous headache. I remember using shutter glasses years ago on a CRT @ around 60hz with plenty of eyestrain/headache as a result.

[DmitryKo]

it sounds like the sensio format that we've been able to see in Avatar the game

I thought Sensio format is just another variation of side-by-side with qunicunx sampling, like the RealD format. Do you have any screenshots to post?

[cybereality]

The Sensio format does appear to be side-by-side checkerboard:

Avatar_SENSIO.jpg

Avatar_SENSIO2.jpg

[DmitryKo]

I would rather have 120hz checkerboard over 60hz full frame

Both have 60 Hz per eye; the eyestrain was due to the some properties of CRT displays, which do not apply to fixed-pixel displays with passive backlight or pulse-width modulation of brightness.

The Sensio format does appear to be side-by-side checkerboard

Well, judging by the HUD, it appears that Sensio uses even/even or odd/odd quincunx sampling, and RealD uses even/odd or odd/even (explanation here). Who wants to file a patent application for a top-bottom format with vertical subsamping and quincunx sampling?

[vrekks]

Bastian I think you would have to draw a 1 pixel diagnal line, not horizontal to get only one eye to see it (like only the black squares on a checkerboard). One benfeit of checkerboard is that it uses less processing power than full res 3d. To be perfectly honest 95% of the time I can't tell the tv is only half res per eye. Like others have said, only thin lines and small text give problems. I watched Avatar in a theatre and did not notice any differences between my half res setup and this dual projector, polarized setup.

[shonofear]

but is it easier to read text as checkerboard format compared to horizontal interlaced format?
just thought it being diagonal would have a different effect on viewing/reading text on screen...
cause horizontal interlaced is really hard for me to read text.

cheers

[Neo42]

I am not getting this 'half the pixels are invisible' idea that people keep talking about. With checkerboard technique there is no information removed, each eye receives the same data but across two frames spaced 1/120th of a second apart.

EDIT: My Mitsu C9 series DLP allegedly has 3 DarkChip4 DMDs which are supposedly natively 1080p panels, which effectively would give this display an advantage of not losing resolution while wobulating for 3D.

[DmitryKo]

With checkerboard technique there is no information removed, each eye receives the same data but across two frames spaced 1/120th of a second apart

No. Each view is half-resolution (quincunx sampled) video at 60 Hz, not 120 Hz; the resulting frame alternative stereo stream is 120 Hz.

give this display an advantage of not losing resolution while wobulating for 3D

Horizontal wobulation is unrelated to frame sequential stereo; the DMD is capable of native 120 Hz.

Check Explain checkerboard and interlaced 3D pls

but is it easier to read text as checkerboard format compared to horizontal interlaced format?

Yes, it should be easier.

Side-by-side checkerboard format is even better, because both left and right views can be sampled at the same positions (as in Sensio and probably RealD), which should improve readability of text placed at screen-depth.

[shonofear]

but is it easier to read text as checkerboard format compared to horizontal interlaced format?

Yes, it should be easier.

Side-by-side checkerboard format is even better, because both left and right views can be sampled at the same positions (as in Sensio and probably RealD), which should improve readability of text placed at screen-depth.[/quote]

Soo...
Question then, maybe be silly but would it be possible to use my simple wired X3D shutter glasses with say the TRIDEF checkerboard output mode and somehow get it to work and be in sync on a regular CRT monitor or a 85Hz DLP projector?

Just looking for a alternate interlaced option where I can actually read text (pageflip not option atm)

Thank you your intelligence,

[DmitryKo]

X3D shutter glasses with say the TRIDEF checkerboard output mode and somehow get it to work and be in sync on a regular CRT monitor or a 85Hz DLP projector?

No, you won't see any stereo. The trick is to decode the 60 Hz checkerboard video at the display/projector and convert it to frame-sequential 120 Hz; it can't work without the support on the display side.

[shonofear]

mmm, I see.
Thanks anyway,

[GoldChain]

No, you won't see any stereo. The trick is to decode the 60 Hz checkerboard video at the display/projector and convert it to frame-sequential 120 Hz; it can't work without the support on the display side.

New to the forum..... lots of background with audio and video formats, etc, but trying to decipher the whole situation with the 3D formats, etc..... as I'd LOVE to A) upgrade from my current TV, and B) get in on 3D. (for the record, I've got a PS3, so I'm hoping to use that to drive the 3D display that I eventually buy).

So, is the checkerboard format indeed at 60Hz? If so is that a 1080i@60Hz? or 1080P@60Hz? Bandwidth-wise, 1080i@60Hz = 1080p@30Hz = Best single eye HD quality (i.e. full resolution). If the checkboard is being transferred at 1080i@60Hz, then it would be the same amount of data as non-stereo video, therefore each eye woudl be getting 1/2 of the resolution (which is what I understand is the case from reading this thread). If it's being transferred as 1080P@60Hz, that's double the bandwidth of full-resolution non-stereo video, in which case both eyes should be able to get full-resolution @ 30fps (which none of the broadcast or BD HD formats use more than 24/30fps, right? (I know technically there's 720P@60Hz, but seriously, why hasn't 720P gone away yet?))

Either way, why would 120Hz be required? As a side question: Why are the folks who make these standards stop screwing around and just straight to full-frame, full-resolution 3D? Encode 2 separate 1080P streams (interleaved onto the media for easy reading), and deliver those in sequential format for the display to , uh, simply display. I'm blown away at the notion that intelligence needs to be built-in to the display. IMHO, the display should have NO part in interpreting any video format, etc. It should simply take a stream of bits and display them. Otherwise, now you have to worry about what "formats" your TV supports in addition to your media players, etc...... Geez....

GC

[DmitryKo]

is that a 1080i@60Hz? or 1080P@60Hz?
If it's being transferred as 1080P@60Hz, that's double the bandwidth of full-resolution non-stereo video, in which case both eyes should be able to get full-resolution @ 30fps)
Either way, why would 120Hz be required?

It's a 1080p60 signal, which contains a stereo picture in each video frame, encoded in a checkerboard pattern. So it has the same bandwidth as 1080p60 full-resolution mono signal, but contains two half-resolution 60 Hz pictures per each eye, or 120 Hz in total.

Why are the folks who make these standards stop screwing around and just straight to full-frame, full-resolution 3D? Encode 2 separate 1080P streams (interleaved onto the media for easy reading), and deliver those in sequential format for the display to , uh, simply display.
IMHO, the display should have NO part in interpreting any video format, etc. It should simply take a stream of bits and display them.

As I said here, you are confusing video compression formats with display link formats.

On the Blu-ray 3D media, the steroscopic movies are encoded with a Multiview Video Coding extension which basically stores a L stream and the difference between two L/R video streams. This is slightly more efficient that just coding two independent L/R video streams, one for each eye (typically 25% less bitrate). This is a device-independent full-resolution format.

The video display link format depents on the display technology used by the stereo 3D display, this is why the player has to encode the video stream to an uncompressed format that is supported by a particular device. That includes checkerboard for DLP displays, "frame packing" top-bottom format for 120 Hz LCD/LED TVs and plasma displays, line interleaved for certain computer monitors, frame alternative format for 120 Hz computer monitors, etc.

A special case are TV broadcasting formats which do not use MVC and involve half-resolution compression formats with top/bottom and side-by-side frame arrangement, which use 1080i60 video signal to fit withing the bandwidth limits of legacy broadcasting platforms. Technically, the set-top-box should be able to upscale them to standard full-resolution formats, however recent HDMI 3D spec includes them as native video link formats, for whatever reasons.

[GoldChain]

Good stuff DmitryKo.

I want to be clear, for sake of discussion. I'm aware of the conceptual differences between compression codecs and media transmission (display link).

Encoded Media
\/ \/ \/ \/
Player reads media and decodes encoded video into a "raw" (uncompressed) video stream
\/ \/ \/ \/
Player take raw video stream and puts into another "format" that the display can handle
\/ \/ \/ \/
Formatted display stream is then transmitted via specific interface type (HDMI, component, etc) to display
\/ \/ \/ \/
Display take video signal/stream in and displays it

This is probably an oversimplification of the process, but pretty close. In the case of 3D, the display is doing more than simply displaying the data. It's buffering and showing the same "frame" of video multiple times (i.e. if it's doing a 3:2 pulldown or displaying stereo content, it's alternating views).

My point was, and it's more of a gripe of the path the manufacturers have chosen, IMHO, I want a "dumb" display: feed it pixels, and it displays them. In other words, there should be a single stream formatting that ALL displays that are 120Hz should be able to accept in order to handle stereo video (i.e. page-flipped or top-bottom frame packed, etc... it doesn't really matter, but everyone should just settle on ONE so that ANY player will work with ANY display).

IMHO, the manufacturers are making this way too complicated for no reason. Putting wifi, etc into TVs is just insane. Let the display be a dumb display and have the "smart" device drive it. That way you're not locked into the limitations of the hardware of the display (i.e. if you have a TV with wifi so you can watch YouTube or Hulu, but then they change the video format they play, it may not work on your TV anymore...... let your multifunction PC/Gaming console handle the heavy lifting, and make the display just WORK).

GC

[DmitryKo]

Unfortunately many legacy requirements preclude from following such a simple path as you describe. The problem is, computer displays and consumer televisions have evolved in very different directions for a very long time, and only recently there have been a strong trend for convergence.

As for 3D formats, 120 Hz frame alternative displays are not the only stereo displays in existence, so there can't be one single "fit it all" format such as top/bottom, "frame packing" or frame alternative (these three are essentially the same from the practical point of view). There are quite a few devices with dual-engine setup, so these are better served by a side-by-side format, and there are quite a few devices built around line-interleaved format... not to mention dual projection setups which require dual video interfaces, which is not even supported by HDMI 3D.


Hopefully VESA DisplayPort 1.2 will someday emerge as the ubiquitous full-resolution stereo interface for both 3D TVs and 3D monitors. I personally look forward to having a 27" 2560x1440 120 Hz monitor, which should only be possible with DisplayPort...

[GoldChain]

Good stuff, man!

I think we're on the same page as far as a future ubiquitous standard. It's funny as we still refer to TVs and computer monitors as if they're separate, which only for the fact we consider them separate could be a single category called "Displays".

That's kind of the point of my original "gripe" was that if a display (PC, TV, whatever) simply accepted a bitstream of video data to display that stream at a given resolution and refresh rate, then you've just removed any role the display would need to play in displaying any content. Then it would simply be the device that streams the content to determine if it's non-3D content, or 3D frame alternate (i.e. at that point there would be no need for any other 3D delivery format, as frame alternate would work great with shutterglasses, etc) then send that stream to the display. Boom - done.

Alas, I realize I can't change history, nor the path displays have taken to get where they are. It seems to me it would be beneficial (not perhaps for manufacturers that was to distinguish their products from competitors) to start fresh with displays and formats so that everyone's on the same page

GC

[tritosine5G]

EDIT: My Mitsu C9 series DLP allegedly has 3 DarkChip4 DMDs which are supposedly natively 1080p panels, which effectively would give this display an advantage of not losing resolution while wobulating for 3D.

lol . The amount of wishful thinking there

What do you think of that checkerboard +plasma craze over nvidia forums?

How's it better than 720p + 16x MSAA ? I don't even ....

Btw its very good now nvidia is forced to innovate something for 720p. We can avait their answer to mlaa and for 3d its particulary important.