Hello ! I'm new here, this isn't exactly my first post, but this is why I signed up ^^.
I was wondering, and couldn't figure out myself, how the resolution of a 3D stereoscopic movie works, with shutter glasses.
Let's say I have a side-by-side 3D movie, which in Avinaptic shows up as 1920x1080. However, of course, it's 2 "compressed" frames side by side. When I watch the movie, are the frames displayed as 960x1080, or the original frames, before being "compressed", were 1920x1080, and thus I'm seeing each frame in Full-HD ?
I mean, what puzzles me is, if it's latter, why doesn't it show up as 3840x1080 ?
Thanks, and let me know if I wasn't clear !
+ Reply to Thread
Results 1 to 24 of 24
In SbS 3D the original L & R full 1920x1080 frames were each reduced to 960x1080 then stitch together into one 1920x1080 frame. In playback, each composite frame is split back into the two 960x1080, each section reinterpolated back to 1920x1080, then sequentially shown to the left and right eyes.
- no increase in total bandwidth requirements
- sequential display frequency can be increased to prevent flickerng
- each eye ultimately sees only half of the original horizontal resolution
- watching in 2D still requires the same decoding & processing complexities and delays
The sophistication of the decoder will determine how good the 960 is reinterpolated back to 1920 on the fly, which is what is seen in playback.
Last edited by turk690; 22nd Mar 2014 at 10:00.For the nth time, with the possible exception of certain Intel processors, I don't have/ever owned anything whose name starts with "i".
Thank you, that answered completely my question, but forgive my ignorance, what exactly does it mean that the sections are "re-interpolated to 1920x1080" ? Is that why the image doesn't appear stretched ? Link me to a wiki or page if you want.
EDIT- Don't get me wrong, it's not that I'm lazy, I googled for it but I was unsure what kind of interpolation it was, since I know it's not exclusive for video decoding (I knew it was used on laser optical mice and such)
Then why each eye sees only half of the original resolution ? This is a little confusing
Sent from my iPod touch using Tapatalk
Think of it like this: Suppose you resized the 1920x1080 frame to 1x1, a single pixel. How much detail is lost? All of it. How much detail can you get back by resizing that single pixel back to 1920x1080? None of it. You'll just see one gigantic pixel.
Oh I see, now I understand, thanks. Would you be so kind to answer just one last question ? Why can't companies squish double-sized frames, so you actually have a full-hd image in the end ? Is it a matter of disc space or the cables don't support such a big stream of data or what else ?
Sent from my iPod touch using Tapatalk
Originally Posted by arator95
Am I mistaken?Donatello - The Shredder? Michelangelo - Maybe all that hardware is for making coleslaw?
Indeed you would need shutter glasses to see a full hd image, but the input should be full hd too. If I understood what the other guys replied, since the frame is not full hd to begin with, it is useless to have shutter glasses unless you use them for video games or you actually make a video where each frame is originally 1920x1080. Correct me if I'm wrong please.
Sent from my iPod touch using Tapatalk
I was going to hold off on this LENGTHY explanation, but I think everybody needs to be on the same page, so here goes:
You must understand that there is a distinction between a STORAGE format, a TRANSMISSION format and a DISPLAY format.
Working backwards, for average consumers there really is only 3 DISPLAY formats (not counting PC-based or Photography-based ones):
1. Anaglyph (aka Red/Cyan, Green/Magenta, Blue/Yellow-Amber, etc.) - lowest common denominator and available to everyone who has access to the glasses, as you can use a standard 2D monitor/screen to view in 3D. Also lowest quality, by far.
2. Alternating LC (liquid crystal) shutter glasses (aka "Active") - requiring a transmitter synced to the image changes & powered (hence "active") glasses on the receiving end. Can be lower quality (if using older 1/2 framerate alternations per eyeview) or higher quality (if using newer/current full-framerate alternations).
3. Polarized glasses (aka "Passive") - requiring a means of generating & retaining images of alternate optical polarization & simple polarizing filters in the glasses. This can be linear polarizing or circular, but although linear has better isolation from crosstalk (ghosting), it is very sensitive to head tilt, so almost all modern systems use circular (aka clockwise vs. counter-clockwise polarities). Can be of medium quality (if using alternate line/column "FPR" polarizing sheets) or high quality (if using a double framerate alternating polarizer). All consumer passive TVs use the medium quality version, but some consumer projectors and ALL professional projectors use the high quality method (known often by the tradename "Z-screen").
1. HDMI 1.4-compliant "Frame packing", which uses 2 full HD images per frame, with special messaging in the metadata that defines the boundaries of each of the 2 images (plus a padding border between). For 1080p images, it gives an actual framesize of 1920x2205 (aka 1080 TaB with a 45pixel padded border between).
2. HDMI 1.4-compliant "Side-by-Side", which uses 1/2 rez horizontal quality (anamorphically squeezed) with full rez vertical quality. Used often in broadcasting.
3. HDMI 1.4-compliant "Top-and-Bottom", which uses full horizontal quality with 1/2 rez vertical quality (anamorphically squeezed). Used somewhat often in broadcasting (though not quite as often as SbS).
4. A host of rarely-used, "optional" formats which are HDMI 1.4-compatible - Field Alternating (analogous to Interlaced), Frame Alternating, Side-by-Side FullWidth, Top&Bottom FullHeight, and 2D+Depth and 2D+Depth+OcclusionBkgd+OcclusionBkgdDepth. Since it isn't mandatory that displays be able to utilize any of these, there are NO known displays that are specifically able to support them.
5. Other, non-HDMI 1.4-compliant transmission channels (which includes HDMI 1.3, etc), all of which are NOT specified in any international & industry-wide standard. This includes PC-related Frame-Alternating (aka Page Flipping) methods which output Full HD at double the framerate to maintain the normal framerate Per Eye.
6. Checkerboard format, a slightly greater than 1/2 rez horizontal & slightly greater than 1/2 rez vertical format that appeared on Consumer displays prior to the emergence of the HDMI standard, which incorporates both horizontal & vertical interlacing.
7. Sisvel (or offshoot version, 3DZ), which incorporates 2 full HD 720p images (plus Depth, in the case of 3DZ) broken up & repacked into the standard 1080p Framesize. Very good compatibility (both 3D and 2D), even though not fully utilizing the highest 1080p quality, but not really implemented yet. Nextgen broadcast channels MAY make use of them, we'll see.
8. Anaglyph. Again, lowest common denominator, again lowest quality
1. MVC, which is what is used with BD3D-compliant discs, as well as a few prosumer cameras. Maintains (nearly) FULL HD for both eyes yet only uses up ~ 1.5x the storage of a single 2D view. I say "nearly" because, like its AVC cousin, it is losing something in the MPG4 compression (even moreso in the dependent view).
2. Dual-stream muxed files, which contain separate & discreet views of each of both eyes multiplexed within a single container file. By design, they are Full HD, and could also be of any of the available compression formats available to monoscopic 2D files (including lossless or uncompressed). Why aren't they utilized more often? Because they ALWAYS use up DOUBLE the bitrate/size that a 2D file would use, and require a demuxer & decoder & player combination that is enhanced enough to fully understand the nature of the multi-stream file and of how to deal with it appropriately.
3. Dual files, each of which are standard 2D files. Has all the overhead of previous format, with the additional burden of needing to keep track of 2 files instead of 1. Only truly helpful when file pairing is guaranteed to be maintained and multiple instances of (non-enhanced) players can be synchronized.
4. Standard 2D file formats that have SbS or TaB or Interlaced or Checkerboard, etc. burned into them. SbS & TaB are commonly found, Interlaced somewhat less (though common in "3D" DVD offerings). All the rest are rare.
5. Anaglyph formats. Again, common denominator and last result.
Each 3D chain makes use of different aspects of these 3 segments.
Blu-ray 3D uses: MVC -> HDMI 1.4 Frame Packing -> whichever display type the user has.
Many of these SbS MKV converted discs/files use: SbS -> HDMI 1.4 SbS -> whichever display type the user has.
Note that, unless you are CAREFUL, it is quite possible to combine a file that is already compromised in the compression with a transmission format that is compromised in the horizontal with a display that is compromised in the vertical, thus leaving you with 3 areas of compromise in your quality. Luckily, even with those 3 areas compromised, the resulting display can look quite decent.
But in order to get the BEST POSSIBLE 3D HD quality, the combinations must use each chain that retains full HD quality in each/both the eye views. This doesn't leave too many options due to bandwidth constraints. 3DBD (Mvc) sent via hdmi 1.4 frame packing to an Active Display is close (only compromised in the MVC compression). Dual file or dual-mux streams played via Stereoscopic player out HDMI/DVI/DisplayPort to a 2xFramerate PageFlipping Projector is another good choice.
So, very long story short: if you were to rip a 3DBD to dual files or dual-mux stream file and played that out of your high powered PC using Stereoscopic Player through HDMI/DVI/Displayport to a 2xFramerate PageFlipping Projector, you'd have all the quality of the original (assuming you used lossless compression on the rips).
Be clear about this: your standard Active TV displays, while they DO 2x Framerate alternating images, they normally DO NOT accept incoming streams that way (though a number of the PC-based projectors will). You would have to give it an HDMI 1.4 FramePacking transmission for it to both automatically turn 3D and retain the full source quality.
So, it can be done, but only in particular circumstances with certain kinds of hardware/software. All other methods and channels are a compromise. But as I said, even the compromise is often worth it.
Last edited by Cornucopia; 25th Mar 2014 at 14:58.
Thanks a lot Cornucopia! That's a lot of info and it helped me a lot. I'm using the shutter glasses from nvidia and a compatible monitor (BENQ XL20 series). I'm going to make a lossless compression of a BD3D, but since the result would be huge, what method would you suggest to lower the bitrate a little, without changing the resolution (keeping the full 3840x1080 or (1920x2160) ?
"Shrek is not Drek !"
Well, what is "huge" to you? And what does "lower a little" mean to you?
I would say, starting from uncompressed, going to lossless compression (huffyuv, lagarith, utvideo) will ~1/2 the bitrate without a loss in quality. To go further, try lighty-compressed lossy formats (cineform, dnxhd, prores). If that is still too large, try h264 with high bitrate. Any lower/smaller and you are bound to start seeing a loss in quality.
1/2 of the bitrate would be already great, thanks a lot man, I'll try this out, and probably the "more lossy" compression too, just to see how far I can push it.
Again thanks you've been great !
"Shrek is not Drek !"
I think there's a misunderstanding here. Compression of a Blu-ray source with a lossless codec will increase the size by a very large amount (~10x). A 90 minute 2d movie, uncompressed to ~400 GB, will likely be in the 200 GB range. The only way to decrease the size would be to compress it more with a lossy codec. I don't know if there are any free encoders that can produce MVC format. That means you'll have to store your videos as two 1920x1080 frames which will double the size (~800 GB) before you start encoding.
Last edited by jagabo; 25th Mar 2014 at 17:22.
Yeah, I hinted at that in the previous bit about storage formats.
And to further clarify on jagabo's, "lossless compression" is only lossless of and ~1/2 the bitrate of a fully decompressed or uncompressed source file. If you were thinking of lossless as being 1/2 the bitrate of the already heavily compressed (with the very efficient 3d mvc codec) Bd3ds, you would be very mistaken. And an attempt to reconvert from mvc to 2 instances of a more traditional (and much less efficient) 2d codec at bitrates that would give you quite a file savings over mvc's, would most certainly lead to MAJOR loss of quality.
You have got to remember, a 3dbd file is HUGE for a reason: there is DOUBLE the HD information being given to you.
Oh snap, there really was a huge misunderstanding. I know I'm going to lose quality but what would be the point where it's not noticeble on a, say, 24" display ? Anywhere near 40-20GB ?
"Shrek is not Drek !"
That often depends on:
1. Your codec choice(s) and their settings
2. How sensitive you are to quality changes
3. The complexity of the material (both spatially & temporaly)
There is no hard & fast rule.
As Cornucopia said, it depends on the particular movie and whether you know what to look for. Losses will be more noticeable on a 24" display viewed from 1m away compared to a 46" display viewed from 4m. You'd have to sit <2m away from a 46" TV to see the same amount of detail (assuming a high quality source and all else being equal).
For most regular 2d movies most people won't notice the difference between the original Blu-ray and an x264 compression to half the size, especially if they weren't viewing the two side-by-side. But for a 3d MVC Blu-ray converted to two non-MVC 1920x1080 frames the differences will be more noticeable. Just try it and see what you get.