Getting the 3D words right


Captions and sub-titling in 3D is not so easy as in 2D. The words cannot simply be superimposed over 3D content. They need to be accurately positioned in the depth plane otherwise they can create visual confusion and break the whole 3D illusion.

For instance, consider a street lamp-post on the right in the foreground and in front of the screen plane. There is a street scene with shops in the far background and behind the screen plane. Fade in a caption with no parallax on the screen plane. This works if the caption is to the left of the lamp-post. However, if the caption is positioned such that it intersects the lamp-post, then there is a 3D confusion. You have the post in front chopped off by the caption behind it. Impossible!
To get the caption or subtitles moved in depth (z-plane) is relatively easy. You need a left and right copy of the same graphics and key. They have to be displaced horizontally by the required number of pixels for the depth positioning. Finding the correct position is not so easy and the action in the scene could be moving objects in and out of the plane all the time. You can’t simply place the captions permanently fully forward. Disparity limits and range have to be considered to find a comfortable position that works.
The recently-approved DVB-3DTV specification is accompanied by a Subtitling Systems specification (EN 300 743). This provides a standard way of allowing subtitles and other on-screen graphics to be positioned in the stereoscopic picture. There are already a number of systems to help address problem when it come to off-line preparation. Using the disparity data gathered from media, these software tools can automatically position captions or graphics and provide sub-titling positioning assistance. To do this on live programme content is more challenging.
Depth disparity analysis needs to be done in real-time to determine the positions of the most forward objects so that a suitably forward position for the captions or graphics can be estimated. However, this data can be restricted to a local rectangular target zone over which these are going to land.

The Cel-Scope3D stereoscopic analyser is able to provide suitable position vectors whilst also checking the 3D content for disparity errors or depth budget problems. The position information then has to be acted on by a 3D shifter box and the CG generating the overlays. Some vision switchers can also make the necessary depth adjustments so that words and action don’t crash in 3D.
Robin Palmer is Managing Director of Cel-Soft and is currently involved with software solutions for 3D & TV quality control and measurement technology.

Tags: iss052 | 3d captions | 3d sub-titling | subtitles | captions | dvb-3dtv | depth disparity | cel-soft | N/A
Contributing Author N/A

Read this article in the tv-bay digital magazine
Article Copyright tv-bay limited. All trademarks recognised.
Reproduction of the content strictly prohibited without written consent.

Related Interviews
  • Screen Systems show SASQUATCH for subtitles at IBC 2018

    Screen Systems show SASQUATCH for subtitles at IBC 2018

  • Cavena Image Products at BVE 2015

    Cavena Image Products at BVE 2015

  • SysMedia at IBC2011

    SysMedia at IBC2011


Articles
BSC Expo 2019 Report
Paul MacKenzie BSC Expo returned to the Battersea Evolution on Friday February 1st and Saturday 2nd. It is a busy and friendly event though this year in need of some temperature control: exhibitors around the entrance area were uncomfortably aware of the wintry conditions outside and the main hall was in need of cooling.
Tags: iss136 | bsc | cinematography | canon | c700 | c200 | cartoni | holdan | blackmagic | peli | panasonic | sennheiser | sony | fs5 | teradek | viten | flowtech | Paul MacKenzie
Contributing Author Paul MacKenzie Click to read or download PDF
Looking for the Silver Lining
Harry Grinling According to the World Meteorological Organisation, there are 10 different types of cloud, each of which can be divided further into sub-types. They range from the cirrus, the thin floaty clouds which generally serve only to make the sky look beautiful to the towering, all-embracing cumulonimbus which can deliver fearful quantities of rain – the biggest cumulonimbus clouds can contain 50 million tonnes of water.
Tags: iss136 | cloud | lto | archive | storage | Harry Grinling
Contributing Author Harry Grinling Click to read or download PDF
TV Futures - The Shadowing Experience
Daniel Jones My name is Daniel Jones, and it is no accident that I’m currently studying BSc (Hons) Television and Broadcasting at the University of Portsmouth. Since completing GSCE media studies I have been constantly questioning what I watch with questions such as, “Wow, how was that filmed?” or “That looks amazing, I wonder how long that took?” It should come as no surprise that I made it a big focus of mine to get myself some real work experience to give myself some answers to these production questions.
Tags: iss136 | portsmouth university | runner | student | education | training | Daniel Jones
Contributing Author Daniel Jones Click to read or download PDF
Streaming - Shut up and Sit Down
Matt Lees Twilight Imperium is a board game all about strategy and backstabbing. Known for its length of gameplay, the team behind Shut Up and Sit Down decided to deliver a mammoth 10-hour live broadcast, streamed to Twitch. Matt Lees is here to tell us how it went down.
Tags: iss136 | streaming | blackmagic | atem | 4k | canon c100 | sony pxw-z90 | Matt Lees
Contributing Author Matt Lees Click to read or download PDF
NAB Intelligence
Bruce Devlin - new It's that time of year again. 06:30 on road bikes on the Las Vegas Strip with a merry bunch of folks who believe they look good in lycra. As we pedal West up the hill towards Red Rock there will be stories of the year just gone and questions about what to look for at the show. I, for one, will be attending the Devoncroft event to check up on who is spending what and where. I will then look forward to the topics that might be buzzing around the halls. I predict there will be many displays of Machine Learning hiding under the banner of Artificial Intelligence. Some of these will show better picture quality, others will show personalisation engines, some will be improved search tools and yet more will be synthesising pictures and motions to replace reality with and Artificial reality that's more compelling for the viewer.
Tags: iss136 | nab | class | ai | smpte | st2110 | imf | Bruce Devlin - new
Contributing Author Bruce Devlin - new Click to read or download PDF