Further 3D Fun

  • A couple of days ago, I mentioned that the shelving unit I had prevously bought to stand my (original 2D) TV on turned out to be too high for my new 3D TV. This is because the tv stand provided with it pivots from side to side, but has no tilt; and viewing the TV from below spoils the 3D effect, with significant crosstalk. However, viewing it from the “sweet spot” gave a perfect picture – unfortunately, my sofa isn’t that high!

    I was contemplating wall-mounting it with a tilt bracket, but today I decided to move some furniture about. This was not a decision lightly made, as I had only just got all my audio-visual separates set up.

    Anyway, the TV is now on a 700mm high unit, with a long floor level-shelf that all my boxes fit on. This is still higher than normal TV units, but brings the vertical angle into the area of tolerance. The 3D effect is now spot on, and is just a matter of hitting a button, rather than jamming books under the back of the stand, and trusting that it won’t topple. (The topple factor possibly being why the tv stand doesn’t tilt.)

  • Many people who wear glasses complain about having to wear 3D glasses on top of them. I find it depends on the kind of 3D glasses. The ones that came with the TV are fairly ok, but the ones my local cinema give you are actually quite excellent. I have a couple of pairs of those, so have been using one of them with my TV.

    However, whereas wearing double-specs while sitting upright in the cinema is one thing, doing so while lounging at home is another. So today I broke up one of the cinema RealD 3D frames, to liberate the lenses. These are just thin polarized plastic film. Taking care to keep the right and left lenses in the correct order, I then took an old pair of prescription glasses – not my latest prescription, but good enough for watching TV – and sellotaped the film, top and bottom, to the lenses.

    You’d think the tape would be a distraction, but I’ve managed to keep it well out of my line of vision, and I don’t notice it at all. So now I have got a pair of prescription RealD (passive 3D) glasses! They work so well, I may take them with me next time I go to see a 3D movie at the cinema. Although now I know it works, I may replace the tape with superglue.

    The Real 3D lenses are smaller than my glasses lenses. This doesn’t affect my eyesight at all (as the area of the lens my eye is looking through is covered), but it does look weird. What I really need is a big pair of Real3D glasses, or a smaller pair of specs.

  • One of the features my new TV has is the ability to turn 2D programs to 3D, on-the-fly. Generally speaking, the reviews of 3D TVs have been disparaging about this kind of feature, but I have found – used on the right material – it is pretty effective. I am not sure I will use it regularly, but tonight, as a novelty, I have watched both X-Men Origins: Wolverine and the final episode of the current series of Dr Who, in fake 3D. To be honest, it was the only think that kept Dr Who interesting (but that is a rant for another day).

    While doing this, I encountered a couple of adverts that took on a whole new life in 3D. There is a Lurpac Butter ad of a guy making an omelette, that looks positively cinematic. Plus anything cartoony looks great too. Hmmm… must find time to try out Sin City.

2 Comments

  1. October 2, 2011
    Reply

    Glad you got it sorted. =:o}

    The concept of mere software being able to turn 2D images into 3D still boggles me. It was hard enough teaching software to convert colour anaglyph 3D to a full-colour 2D image- A situation where you’re staring with all the information you need, and just have to move it to the right place/colour. Generating depth information from the scant clues available in 2D, without the interpretive power of an actual brain… =8oO

    I wanna know how they do it. I mean, I know the basic set of clues they work from (detecting moving occlusions etc.), and there’s gotta be some powerful and very thoroughly trained neural network stuff going on there, but I wanna know *exactly* they how they stitch it all together. And then I wanna apply that to frame interpolation/extrapolation, because that would be just about unbeatable…

    But alas, it seems the key techniques (many of them pioneered by the good folks at MSU) seem to be all tied up in corporate licensing deals, with only very basic early versions released to the VirtualDub/AVIsynth communities who provided the tools for the research in the first place… =:o

    I had a chance to correspond a while ago with someone at MSU who was looking for a topic for his research/development project, which was roughly in this area. I’m hoping my suggestions as to how to feed something useful back to the community (i.e. openly publishing the data format(s) for carrying the motion analysis data, so that we’ll be able to develop our own plugins that interface with – or at least can crudely emulate – whatever the MSU puts out), while retaining the all-important patents & code for the specific *techniques* that generate the data for the university, fell on fertile ground… but I haven’t heard anything back yet.

    [CROSSES FINGERS]

    [ALSO CROSSES EYEBALLS, THEN TRIES TO FIND A CROSS-EYE 3D CLIP TO VIEW]

    [BELATEDLY REALISES THESE TWO STEPS ARE IN THE WRONG ORDER]

    • chris
      October 2, 2011
      Reply

      I know, it boggles me, too.

      I don’t think you need me to tell you that it is not a perfect algorithm, and sometimes it just looks wrong. Watching J J Abrams Star Trek reboot tonight on TV (I had seen it before, but turned it on to see what 2D->3D did to it), it was a mess – the repeated use of (digital) camera flares screwed up the system.

      However, there are plenty of programmes where it does work, and while the 3D can sometimes look a little bit like moving cardboard cutouts, I have seen plenty worse in studio-converted films.

      There is no doubt that the real benefit of a 3D TV is watching proper 3D content. But with that few and far between at the moment, a lot of fun can be had with 2D->3D.

Leave a Reply

Your email address will not be published. Required fields are marked *