Everything you ever wanted to know about 3D but were afraid to ask

Steve Withers

Reviewer
Last week I was lucky enough to be invited by Sony to a special 3D event held at their 3D training facility in Basingstoke. This is one of four facilities that Sony currently has in operation, the others being at Culver City to provide support to Hollywood productions, Mumbai to provide support to the Bollywood film industry and Hong Kong which supports Asian filmmakers. At these facilities Sony run a series of courses that teach industry professionals about the history of 3D, the development of the technology for both capture and delivery and a guide to the correct 3D filming techniques. The courses themselves are run by Sony Professional Education And Knowledge (SPEAK) and whilst they cover 3D techniques for both the cinema and for TV the emphasis tends to be on filmmaking at Culver City, Mumbai and Hong Kong, with Basingstoke concentrating more on 3D broadcasts.

The course is run by Paul Cameron and what he essentially did was condense the first day of the course into three hours for the benefit of the assembled journalists. During Paul’s presentation we covered everything from the history of 3D to the development of modern 3D capture and delivery technology to the dos and don’ts of creating good 3D. After all that we actually got a chance to put our new found knowledge into practice and shot some 3D footage using a Sony professional 3D rig. Before I go into more detail about areas that Paul covered I would just like to thank him and Sony for what was an absolutely fascinating and very enjoyable day.

245098d1307269624-everything-you-ever-wanted-know-about-3d-but-were-afraid-ask-sony3.jpg


History of 3D

Believe it or not, the first 3D movie was actually created as far back as 1903 but the first 3D boom really took off during the period 1922-25. This first boom in 3D was in fact the only time that anaglyph glasses were actually used for theatrical presentation because the traditional popular image of rows of cinema goers sat in their red and cyan glasses is largely a fallacy. However the 1920s boom was the exception and unfortunately because they used anaglyph glasses the colours were quite bad, in addition the technology was obviously quite primitive. Creating a camera rig with two synchronised cameras was hard enough but synching the projectors proved even more difficult. Given that the art of 2D filmmaking was still in its infancy there was obviously little understanding of 3D filming techniques and as a result of this and the technical limitations the boom quickly died out.

The evolution of 3D would take a big leap in the 1930s with the development by Edwin Land of film polarising technology. The first use of this technology for 3D was at the World’s Fair in New York in 1939. It was here that Chrysler used linear polarised glasses for the first time to present four short 3D films. Paul actually has a very rare pair of these glasses in his personal collection which he was kind enough to show us.

245096d1307269598-everything-you-ever-wanted-know-about-3d-but-were-afraid-ask-sony1.jpg


The second 3D boom was in the 1950s when studios were trying to compete with television and get audiences back into the cinema by offering an experience that they couldn’t get at home. This led not only to the 3D boom but also the introduction of widescreen formats like CinemaScope which of course are still with us today in one form or another. Thanks to linear polarised glasses the colours were now much better but it was still difficult to synchronise the projectors and the camera rigs were very ungainly. In addition the production values were often limited and there was still little understanding of 3D filming techniques which resulted in quite poor content. It is interesting to note that even Alfred Hitchcock dabbled in 3D, shooting Dial M For Murder in the format but unfortunately by the time the film was released the fad had passed and sadly it was never shown theatrically in 3D - now that’s a 3D Blu-ray I’d kill for!

245097d1307269611-everything-you-ever-wanted-know-about-3d-but-were-afraid-ask-sony2.jpg


The third 3D boom was in the 1980s and is probably fondly remembered by many readers at AVForums. I can certainly remember seeing a lot of threequals in the format, including Amityville 3D, Jaws 3D and Friday the 13th - Part 3D as well as forgotten ‘classics’ like Spacehunter: Adventures in the Forbidden Zone. Once again linear polarised glasses were being used but the camera rigs and projector synchronisation was much better although filmmakers were still developing the language of 3D cinema. I can certainly remember a lot of poking you in the eye shots at the time including a swordfish flying at my face at one point in Amityville 3D.

The next big step in 3D evolution came with the development of active shutter glasses in the early 1990s. These were developed as an alternative to linear polarised glasses and were often used during this period by IMAX for 3D presentations. By the mid 2000s the technological problems of 3D filmmaking had all been overcome, thanks in a large part to advances in digital image capture and projection. This resulted in the latest boom - driven in part by the success of films like Avatar - and whilst 3D booms seem to come and go about every 30 years if you add up all the available content (3D broadcasts, 3D movies and 3D games) this is by far the biggest boom in 3D ever.

The reasons for the failure of previous 3D booms are many but they are largely related to the limitations of the technology, poor production standards and excessive use of 3D. The current boom doesn’t suffer from these technological limitations and benefits from much better production standards and a more artistic approach to the use of 3D. In addition the current 3D boom has something that the previous booms didn’t have - an effective way of delivering 3D content into the home. For 3D to survive - especially as a domestic format - it needs to overcome a series of basic challenges many of which will be familiar to anyone who has read our 3D TV reviews. First of all the 3D image needs to be bright enough, this is true both at the cinema and domestically, and for domestic viewing there also needs to be a wide enough viewing angle and no drawbacks when watching 2D material. For both capture and distribution 3D needs to be reasonably cost effective and finally and most importantly there needs to be good content. It is this final challenge - the creation of good 3D content - that the Sony course seeks to address.


Seeing in 3D

When we look at the world around us we obviously see it in three dimensions but interestingly of the eight depth cues that the brain uses to estimate the relative distance of objects, the first six are equally as applicable to content that is in 2D. The eight depth cues that the brain uses are:

Focus and Depth of Field: Focus is an important tool that the brain uses to determine how far away something is, when the eyes refocus the brain interprets that as depth information. Our brains remember how we focus and build up a memory of the relative distance of each object in the scene. When shooting in 2D cinematographers use depth of field to ‘pull’ objects out of a scene, a person might be in focus but the background out of focus for example. When shooting in 3D there are additional depth cues so you can open up the depth of field much more but in doing so you might lose some that cinematic look.

Perspective and Size: The brain knows that things get smaller as they get further away and is always searching for the vanishing point in every scene we look at. This is the point, often on the horizon, where things become so small that they disappear all together. which results in may compositions having a vanishing point. Straight lines and the relative size of objects help to build a map in our minds of the relative distance of objects in the scene.

Occlusion: Certain objects will hide (occlude) other objects within a scene and the brain makes certain assumptions based on accumulated experiences as you grow up. If a shape appears broken by another object we assume the broken object is further away and behind the object causing the breakage.

Lighting and Shadow: Light changes the brightness of objects depending on their angle relative to the light source. Objects will appear brighter on the side facing the light source and darker on the side facing away from the light source. Objects also produce shadows which darken other objects. Our brains can build a map of the shape and relative position of the objects in the scene from the way light falls on them and the pattern of the shadows caused.

Colour Intensity and Contrast: Even on the clearest day objects appear t lose their intensity the further away they are in a scene. Contrast (the difference between light and dark) is also reduced in distant objects. We can build a map in our minds of the relative distance of objects from their colour intensity and level of contrast.

Relative Movement: As we walk through a scene, close objects appear to be moving past us faster than distant objects. Watch an aircraft landing from a distance and it almost appears to be standing still. Watch it from the runway and you get a very different impression. The relative movement of each object compared to others provides a cue to their relative distance. Cartoonists use this movement to give and impression of depth in animation and film and television producers often use relative movement to enhance depth.

Vergence: Vergence is a general term for both convergence and divergence. If we look at objects in the far distance both our eyes are pointing forwards, parallel to each other. If we focus on an object close up, our eyes converge together. The closer the object the more the convergence. Our brains can calculate how far away an object is from the amount of convergence our eyes need to apply to converge on the object. Film and television producers can use divergence as a trick to give the illusion that objects are further away, but this should be used sparingly because divergence is not a natural eye movement and may cause eye strain.

Stereopsis: Stereopsis results from binocular vision, it is the small differences in everything we look at between the left and right eyes. Our brain calculates which objects are close and which ones are further away from these differences.
 

Attachments

  • Sony1.jpg
    Sony1.jpg
    161.3 KB · Views: 7,347
  • Sony2.jpg
    Sony2.jpg
    164.1 KB · Views: 6,626
  • Sony3.jpg
    Sony3.jpg
    156.7 KB · Views: 6,432
Last edited:

Steve Withers

Reviewer
Working in 3D

All 3D material is based on the concept of creating two different views - one for the right eye view and one for the left eye. In doing so the content creators are taking advantage of stereopsis and vergence to create images that have genuine depth when viewed in the appropriate way. The capture of these two images can either be done using a 3D camera or it can be created in a computer using CGI.

For live action 3D you will need a camera that can capture the left and right eye images simultaneously and to do this you need two lenses to mimic our two eyes. The distance between the centre of both eyes is called inter-occular distance which for humans is about 65-75mm. The exact same approach is taken with 3D cameras only it is called the inter-axial distance, which is the distance between the centre of both lenses. The inter-axial distance needed on 3D cameras is often much less than with human eyes - 15-20mm - which can be a problem with side-by-side rigs. As the name suggests a side-by-side rig uses two cameras next to each other but this results in a wider inter-axial distance which makes close up work difficult. To get around this problem you need to use smaller lenses which will affect the quality of the images.

245101d1307269783-everything-you-ever-wanted-know-about-3d-but-were-afraid-ask-sony5.jpg


In order to shoot close up but still retain image quality a professional 3D camera rig will rely on the use of mirrors. There are three different types of camera rig that use mirrors, the opposing rig, the bottom mount or the top mount. The opposing rig is an older design in which the cameras face one another and use two mirrors to create the left and right eye images. The benefit of this approach is that each camera receives 100% of the light but the rigs are large and unwieldy and for that reason they are no longer used. Hitchcock actually used an opposing rig when he shot Dial M For Murder. The other two types of mirror rig are essentially variations on a theme, one has a second camera above facing down whilst the other has a second camera at the bottom facing up. The advantage of this system is that you can can turn the inter-axial distance down to zero so they are good for close up work. The problem with this approach is that only half the light gets to each camera and the alignment with the camera above is prone to interference from dirt, rain and sunlight. The alignment with the camera below doesn't have those problems but can have difficulty pointing down. Sony used a bottom mount 3D camera rig last year at the FIFA World Cup in South Africa.

In 3D two images are projected on onto the display. By wearing a special pair of glasses the two images are split so that each eye only sees one of the two images. When comparing the left and right eye images, every object in the scene is horizontally displaced by a small amount. The brain assumes these two displaced objects are actually one object and tries to fuse them together. The only way it can do this is to assume that the object is either in front of or behind the screen plane. The direction and the amount of displacement (disparity) defines where each object is in the 3D space. If the object is displaced slightly to the left of the left eye and to the right of the right eye then the brain will assume that this is only one object behind the screen - this is called positive parallax. Conversely if the object is displaced slightly to the right of the left eye and to the left of the right eye then the brain will assume that this is only one object in front of the screen - this is called negative parallax. However if the object for the left eye and right are in the same position on the display the brain sees this as one object on the screen plane with no 3D displacement - this is called zero parallax.

245102d1307269796-everything-you-ever-wanted-know-about-3d-but-were-afraid-ask-sony6.jpg


When shooting in 3D there are a number of ways of manipulating the image. When the inter-axial distance is increased then the perceived depth increases and the distance between objects in the scene appears to increase. If the inter-axial distance is increased this will enhance the 3D which is called hyperstereo and if the inter-axial distance is reduced this will crush the sense of 3D and this is called hypostereo. However care should be taken not to increase the inter-axial distance too much because close objects will appear closer but no larger and will break the rules established by the brain's understanding of depth cues. Alternatively when the angle between the two cameras is altered to point them slightly towards each other, commonly referred to as convergence angle, the perceived 3D image goes further into the distance, even though the various objects appear to be separated by the same distance. Care shold be taken not to push the 3D scene back too far as objects in the distance may force eyes to diverge which may cause eye strain. Excessive convergence angles will also introduce keystone errors that will need to be corrected later.

245103d1307269813-everything-you-ever-wanted-know-about-3d-but-were-afraid-ask-sony8.jpg


One of the problems with 3D material is that we are always focused on the screen but the convergence is constantly changing and this split between focus and convergence is called accommodation. Some people are better at accommodation than others such as when looking at autostereograms (magic eye images) which require the viewer to converge on infinity. If you are good at looking at these autostereograms then you will find it easier watching 3D whilst those that find these images difficult to see might suffer from fatigue and eye strain.

The development of 3D has resulted in a new role called the convergence puller or stereographer. Convergence relates to where within the screen plane the images appears, and using convergence the stereographer can move the image from negative to zero to positive parallax. When shooting in 3D the stereographer will use a depth budget within which he will try to keep the majority of the objects in order to make the viewing more comfortable. You don't want an object with too much negative or positive parallax as both can take you out of the experience or prove uncomfortable. The stereographer also needs to be careful of object violations which is when something that should be behind an object appears to be in front of it. For example the scores on football matches are usually placed in negative parallax to try and avoid object violations but if a footballer runs right up to the 3D camera this might still happen. Poor use of convergence can change the depth budget and create bad 3D which was often the problem with 3D films in the 1950s and 1980s.

The most common error when working in 3D is misalignment of the cameras in the camera rig, both cameras must be set at the appropriate inter-axial distance for the scene. They must also be perfectly aligned with one another so that the two images can be mapped on top of one another on the display to provide a good 3D image. This means that the cameras must be at the same height and tilt. It is also important to match the lenses so that they track each other exactly and of course to match the actual cameras in the rig.

When authoring 3D content for screen size, going from cinema to TV is OK but going from TV to cinema can result in images that cause the eyes to diverge due to too much positive parallax. To avoid this problem SBkyB uses a conservative depth budget which allows them to project on larger screens. Another problem to be aware of is ‘3D shock' which is when a film goes from one scene with a lot of 3D depth to another scene with a lot of 3D depth. To avoid this a filmmaker needs to insert a flat 3D image between the two scenes to ease the transition, you can clearly see this being done in Avatar which is a masterclass in 3D filmmaking. For live broadcasts this problem can be avoided by making sure that all the cameras are well balanced. Unlike in the early days of 3D the effect needs to be used sensibly to create a believably environment in much the same way as surround sound is mixed with plenty of immersion and atmosphere and occasional wow moments.
 

Attachments

  • Sony5.jpg
    Sony5.jpg
    184.4 KB · Views: 4,931
  • Sony6.jpg
    Sony6.jpg
    139 KB · Views: 4,784
  • Sony8.jpg
    Sony8.jpg
    155.7 KB · Views: 4,725
Last edited:

Steve Withers

Reviewer
Watching in 3D

As I'm sure most people know, you currently need to wear glasses to watch 3D and whilst there are glasses-free 3D being developed they are still a number of years away. So for the time being glasses are needed for 3D and there are four primary types - anaglyph, polarised (linear or circular), active shutter or wavelength multiplex.

The first and oldest approach is anaglyph which uses a filter on each eye that is the opposite colour to the filter itself, therefore red/cyan, green/magenta and blue/yellow. The combination of green and magenta is the most comfortable on the eyes but red and cyan is easier to print so that tends to be the most commonly used. Contrary to popular opinion anaglyph glasses do not use red and blue. The anaglyph method was used during the 1920s but due to the poor colours it creates and its general inefficiency the technology was dropped after the invention of polarised lenses. Aside from some misguided attempts at broadcast 3D in the 1980s, anaglyph is now used almost exclusively within print media these days.

With the invention of polarised filters the use of linear polarised glasses became the primary method for delivering 3D. It was during the 3D booms of 1950s and the 1980s that linear (vertical or horizontal) polarised glasses were used, although some cinemas in the 50s used glasses which were polarised at a 45 degree angle so that it didn't matter which way around you wore the cheap cardboard glasses. Unlike with anaglyph the colours were much better (in fact the best of any system) but this approach required a silver screen which could be expensive and it reduced the viewing angle and thus the number of seats that the theatre could sell. On the plus side the glasses themselves were very cheap but as anyone who has been to the cinema knows, the brightness was reduced considerably. Another problem with linear polarised glasses was that you needed to keep your head level, otherwise the 3D effect is lost and unlike anaglyph, the polarised system can not be used for printing. These days linear polarised is only really used with IMAX screens, most other polarised systems use circular polarisation.

245104d1307269897-everything-you-ever-wanted-know-about-3d-but-were-afraid-ask-sony7.jpg


The circular polarised system uses a quarter wave plate over the linear polarised film which means the glasses are more tolerant to tilting the head as well as extreme viewing angles. Otherwise this approach has all the same strengths and weaknesses as linear polarised glasses such as better colours and cheap glasses but the cinema will need to install a silver screen. This is the method used by RealD in the cinema and by LG on their Cinema 3D displays which is why RealD glasses can be used with LG passive displays. Both the linear and the circular polarised systems are sometimes referred to as passive or passive polarised. When this system is used on domestic displays a polarised filter is placed in front of the screen which will block out alternate lines for each eye but when combined creates a three dimensional image. This approach has all the advantages of the passive system such as simpler cheaper glasses, no flicker and improved viewing angles but the downside is a loss of resolution and the possible compromise of 2D performance by the polarised filter.

During the 1990s XpanD developed the use of active shutter LCD glasses, initially for the cinema but more recently for domestic displays. The advantage of this approach is that no special screen or projector is required and in the case of domestic displays there is no filter over the screen which means you don't compromise the 2D performance. However active shutter glasses are expensive which is a problem either for cinema owners who need to get them back from patrons or for TV owners who have to buy them. Active shutter glasses also require batteries and a transmitter, need to be cleaned at the cinema, are more prone to crosstalk due to the fast reaction times required and only let in half as much light as a pair of passive glasses. This approach is quite rare in cinemas in the UK but it is obviously the main method used on domestic 3D displays.

The final method of watching 3D involves wavelength multiplex visualisation and this is the approach used by both Dolby and Infitec. This system filters each image into a narrow band of primary colours and uses a filter wheel placed in front of the projector lens. The cost of installation is less than with the circular polarised approach but the lenses themselves are very expensive but easier to clean than active shutter glasses. This system has the best separation but colours can appear a bit ‘thin'. Interestingly Vue cinemas started using this system because it was cheaper to install than circular polarised but high running costs due to lost glasses meant that they eventually moved to circular polarised instead. For domestic displays this system doesn't work with TVs but Sim2 has utilised it with some of their 3D projectors.

For a cinema to meet the DCI specifications for 3D projection they must have a resolution of 2k which means that you either use a 2k projector with frame sequential, two projectors or a 4k projector. Sony make a 4k projector that can project passive 3D using a special adaptor for the lens. The projector uses top-over-bottom resulting in half the 4k frame for one eye and half for the other which when combined create a 2k 3D image which meets the DCI specs for 3D. We were luckily enough to spend some time in Sony's very impressive screening room watching 3D footage from the last FIFA World Cup. This footage was projected using the top-over bottom delivery method and a 4k projector to create some incredible passive images. The footage from the 2010 FIFA World Cup in South Africa was really impressive and gave a much greater sense of size and dimension to the stadiums as well as a greater appreciation of where the players were respective to one another. There was also some footage taken from the Swindon dog racing track which I found surprisingly exciting and just goes to show the potential of 3D when it is done well.

There are several methods for delivering 3D in the home, including digital off-air, satellite, cable transmission, the internet and Blu-ray. At the moment there are no special compression standards designed for 3D, therefore existing standards must be adapted. The left and right signals need to be combined into one HD frame sequence and sent over a normal transmission system. There are three basic ways of delivering 3D - side-by-side, top-over-bottom and frame sequential. The side-by-side and top-over-bottom approaches are good for normal broadcast transmission because the bandwidth and frame rate are the same. Unfortunately the downside of these methods is that the horizontal resolution is halved for side-by-side and the vertical resolution is halved for top-over-bottom. The frame sequential method is good for local connections such as Blu-ray players and game consoles and the resolution is full HD but the bandwidth and frame rate must be doubled to reduce flicker.

In the home the two differing types of display offer differing levels of resolution with active shutter being capable of full 1080p to each eye whilst the passive approach essentially provides 540p to each. However the story isn't as simple as that because whilst active shutter displays can produce the full 1080p resolution with Blu-rays they are only displaying half the vertical resolution if the 3D content is delivered side-by-side. Whereas for passive displays they are essentially displaying half the horizontal resolution for Blu-rays and both half the horizontal resolution and half the vertical resolution with content delivered using side-by-side. This might sound like a huge loss in resolution but as we have found in our reviews of passive displays this is not the case as the additional depth cues trick the brain into thinking he image is more detailed than it actually is. As a side note, 3D content is produced using passive displays because all professional 3D monitors (including the Sony Professional monitor we used on this course) are passive displays. This is because there will be multiple displays in the studio or control room and active shutters are only linked to one display whereas the content creators will need to look at these multiple displays simultaneously.

The current 3D boom at the cinema started in 2007 but obviously it was the release and subsequent phenomenal success of Avatar that really opened the 3D floodgates. Whilst there have been a number of live action movies shot with 3D cameras such as Resident Evil: Afterlife and Drive Angry, the majority of 3D content consists of CGI animation. This is primarily because this form of animation easily lends itself to 3D and the process of creating both 2D and 3D versions is relatively straight forward. The popularity of 3D has also been boosted by the development of domestic 3D displays and Blu-ray as a high quality delivery system. There has also been a great deal of development in the area of 3D gaming, especially for Sony's Playstation 3.

The development of 3D broadcasting has largely been spearheaded by BSkyB in the UK and they launched a dedicated 3D channel last year. This channel provides a range of 3D content including sport such Premiership matches, concerts, ballet, documentaries and movies. The first major sporting 3D broadcast was the 2010 FIFA World Cup from South Africa during which Sony helped filmed and broadcast 25 live matches from 10 stadiums. The broadcasters used Sony's HDC 3D camera rig which is able to capture stunning 3D through its multi-image processor and 3D outside broadcast truck. Plans are already being made for the broadcasting of matches from the 2014 FIFA World Cup in Brazil. Another major sporting event to be broadcast in 3D was the Ryder Cup where Sony worked in conjunction with BSkyB using 24 3D cameras to cover the entire course. The Ryder Cup was a huge undertaking with both Telegenic outside broadcast trucks being used, 12 3ality 3D camera rigs, 4 Element Technica 3D camera rigs and 4 wireless 3D camera rigs. There were a number of smaller remote controlled camera on the course which due to their nature couldn't be 3D so interestingly BSkyB used 2D to 3D conversion for this footage.

The most exciting development in terms of sporting 3D is the announcement that the last three days of Wimbledon will be broadcast in the format. Sony has been working in conjunction with the BBC to install 3D cameras on the Centre Court in order to film the Men's Semi Finals and the Men's and Women's Finals. The BBC considered broadcasting these matches in 3D over Freeview but decided it was too soon, however in conjunction with Sony they will be showing them in selected cinemas.
 

Attachments

  • Sony7.jpg
    Sony7.jpg
    122.1 KB · Views: 5,205
  • Sony9.jpg
    Sony9.jpg
    147.5 KB · Views: 217
  • Sony10.jpg
    Sony10.jpg
    167.3 KB · Views: 236
Last edited:

Steve Withers

Reviewer
Future of 3D?

When you look at the history of 3D there has been a boom about every 30 years - the 1920s, the 1950s, the 1980s and now - and these booms have usually lasted about 5 years before the interested faded and producers stopped creating content in 3D. This current boom has a number of differences from previous ones - the technology is much better, the production values are far higher, the content is more varied and producers have developed a greater understanding of the grammar of 3D filmmaking. However perhaps the biggest difference is that we now have established methods for delivering high quality 3D into the home and there is an enormous amount of support for the format amongst both the content creators and the display manufacturers.

Unfortunately whilst audiences still seem excited by the prospect of large 3D event pictures there appears to be little mass market acceptance for 3D in the home. This lack of interest in domestic 3D is now combining with two worrying trends that have developed over the last year - the use of 2D to 3D conversion for movies and a falling percentage of 3D attendance when compared to those going to see the same film in 2D. Hollywood has never been slow to shoot itself in the foot and once the producers of Clash of the Titans saw the box office numbers from Avatar they quickly converted their movie into 3D during its post-production. The resulting film was an utter disgrace in terms of 3D but it did well at the box office and thus numerous other films were converted in a similar fashion. Even if the conversion process could create believable 3D - which it probably can't - the film would still have been created using the grammar of 2D filmmaking which for all the reasons that I have discussed in this article don't apply to 3D filmmaking. This lack of genuine 3D content is worrying and even more so when films that were actually shot with 3D cameras don't follow the basic rules for how the brain perceives 3D - Tron: Legacy is a good example of this.

The second worrying trend is that fewer and fewer people are going to see films in 3D, preferring instead to see the 2D version. When Avatar was released the ratio of 3D to 2D attendance was 90:10 and that 10% probably equates the percentage of the population that can't see 3D. By the time Despicable Me was released last year the ratio had changed to 50:50 and with Pirates of the Caribbean: On Stranger Tides the ratio has dropped to 40:60. The reason for this might be that the novelty of 3D has worn off or that in these difficult economic times people aren't prepared to pay the inflated ticket prices for 3D. Given that a lot of 3D content consists of films aimed at children, the parents might also object to having to pay much more for an experience where the kids will spend half the time taking the glasses off. It might also be that people aren't fooled by the 2D to 3D conversion process and even with content created in 3D the process might not add value - one complaint regarding Pirates of the Caribbean: On Stranger Tides is that the film was too dark and in 3D people couldn't see what was going on. This is a shame because 3D has a lot to offer and when done well it can really enhance your viewing experience. One good example is Werner Herzog's Cave of Forgotten Dreams where the 3D allows you to see how the contours of the rock actually enhance the cave paintings.

How this will eventually play out is anyone's guess but it is interesting to note that The Hobbit is being filmed in 3D with the first part slated for release in December 2012. By then the 3D boom will have lasted for about five years and given the current consumer apathy and falling 3D attendances it is possible that public interest in 3D will be over. Could it be that The Hobbit will be the Dial M For Murder of our generation?
 

Mr.D

Distinguished Member
Having worked with both stereoscopic and converted 3d films I have to say that the 2d to 3d conversions have improved considerably in the last 2 years.

The hassle of dealing with stereoscopic footage is quite significant in lots of esoteric ways that Phil doesn't mention and that are probably beyond the scope of this thread.

Judging by the amount of eminent film directors who are turning to 2d to 3d conversion for some of their older films and consequently being blown away with the results I now have some hope for the 2d to 3d conversion. This is after being somewhat skeptical myself. Clash was a special case to a certain extent in that the studio wanted the conversion done come hell or high water in a restricted time frame. Most film conversions I've been aware of going on are taking considerably longer 6 months or longer with thousands of operators working on them.

However the latest results prove that the process can produce imagery that is almost indistinguishable from true stereo and often without the problems associateed with occlusion , lens disparity , distortion and color , mismatching reflections and flare all of which require intensive manual correction. True stereo is a significant "pain in the ass" for everyone involved and I fully expect it to go the way of the dinosaur in about 3 years, Conversions may even result in better technical imagery.

If we are now at the stage where big name directors who were extremely vocal about the Clash incident are turning to conversion for 30% of the content in their 3d films and coming to the conclusion that it looks pretty much the same as true stereo I'm confident the conversion technique will not be villified for too much longer.
 

Gordon @ Convergent AV

Distinguished Member
AVForums Sponsor
Interesting article and from my experience I agree that part of the problem with 3D appears to be the inability of programme makers to take in to account how we actually perceive things. Most 3d content I've come across has made me feel ill, presumably as my brain is saying, "fail". As you have already pointed out, some folk are more able to accept 3D than others and I wonder if that may have a significant bearing on whether it actually will be a success after all...
 

DOBLY

Well-known Member
Watching in 3D

The most exciting development in terms of sporting 3D is the announcement that the last three days of Wimbledon will be broadcast in the format. Sony has been working in conjunction with the BBC to install 3D cameras on the Centre Court in order to film the Men’s Semi Finals and the Men’s and Women’s Finals. The BBC considered broadcasting these matches in 3D over Freeview but decided it was too soon, however in conjunction with Sony they will be showing them in selected cinemas.

Interestingly, the BBC have just announced that they will be broadcasting the Finals in 3D.
 

Louis Mazzini

Active Member
Having worked with both stereoscopic and converted 3d films I have to say that the 2d to 3d conversions have improved considerably in the last 2 years.

The hassle of dealing with stereoscopic footage is quite significant in lots of esoteric ways that Phil doesn't mention and that are probably beyond the scope of this thread.

Judging by the amount of eminent film directors who are turning to 2d to 3d conversion for some of their older films and consequently being blown away with the results I now have some hope for the 2d to 3d conversion. This is after being somewhat skeptical myself. Clash was a special case to a certain extent in that the studio wanted the conversion done come hell or high water in a restricted time frame. Most film conversions I've been aware of going on are taking considerably longer 6 months or longer with thousands of operators working on them.

However the latest results prove that the process can produce imagery that is almost indistinguishable from true stereo and often without the problems associateed with occlusion , lens disparity , distortion and color , mismatching reflections and flare all of which require intensive manual correction. True stereo is a significant "pain in the ass" for everyone involved and I fully expect it to go the way of the dinosaur in about 3 years, Conversions may even result in better technical imagery.

If we are now at the stage where big name directors who were extremely vocal about the Clash incident are turning to conversion for 30% of the content in their 3d films and coming to the conclusion that it looks pretty much the same as true stereo I'm confident the conversion technique will not be villified for too much longer.

I’d be keen to hear if you think that any of the 3D conversions released so far are of an impressive/acceptable quality. The little that I’ve seen of them (Clash and a few clips of some others) has been quite as flat and unconvincing (and, frankly, horrible) as Steve mentions in his excellent article here. Which film would you point me towards as being the current state of the art in 2D to 3D conversion?

Steve raised some of the complexities of depth composition during a film's production: All of the high-profile conversions that I know of in the pipeline (the Star Wars films, Titanic, Lord of the Rings*) were all shot and edited with no possible consideration for future conversion to 3D. It’ll be interesting to see how these end up looking since they don’t (not intentionally, anyway) follow, as Steve puts it, the grammar of 3D film-making.

My assumption/bet is that The Hobbit will have significantly superior (native) 3D to the LOTR conversions (unless Steve’s worrying prediction comes to pass and we never actually get to see old Smaug in 3D…). Whilst certain big name directors are indeed making statements about how good their conversions are looking (“They would, wouldn’t they” springs to mind…), I’m going to remain as sceptical as you used to be for the time being. I’d imagine that Potter Part 7 Part 2 ought to be among the best of the conversions to date, so it has a lot riding on its shoulders from my point of view.

You clearly have knowledge and experience in this field, so I hope you’re right, but as for the notion of a good conversion being almost indistinguishable from true stereo in future, I’d maybe begin to agree with you if Tron Legacy is the benchmark for true stereo. If, on the other hand, the likes of Avatar and The Hole are considered the benchmark, then I’d be amazed if a conversion can get anywhere close. Adding layers of depth between objects may be quite achievable, but adding convincing solidity to the objects themselves has been totally unconvincing from what I’ve seen so far.

Speaking of Tron, could one of the reasons that I found its 3D to be so dramatically underwhelming for the most part be that it was an example of what you refer to as 30% of the content in 3D films being converted footage? (In fact, when I saw it at the cinema, I was actually fairly sure that much of Tron was a 2D to 3D conversion.) Are there any known examples of this? Perhaps I’ve already been unwittingly fooled by this in some of the 3D films I’ve seen and been impressed with..?

*Several years ago I made a quick & simple and reasonably convincing (hyperstereo) 3D picture of the Golden Hall at Edoras using a couple of frame grabs taken from a helicopter fly-past shot in The Two Towers: I’m quite looking forward to seeing how that sequence looks in the conversion…
 
Last edited:

Mr.D

Distinguished Member
My understanding of Tron Legacy is that the 3d section of the film is true stereo with no conversion.

However even with films shot stereoscopic not every single shot will have a glaringly obvious sense of depth to it. There is not a lot you can do about this unless you want every shot in a film to be a cheesy rollercoaster POV.

More often than not with conversion a lot of the shots that lack depth would have lacked depth even if shot stereo. Its not necessarily the conversion that cannot provide depth just the nature of the shot itself.

The film has already been shot so its not as if you can go back and reshoot it differently. My own take on this is that shots that will obviously not have any real sense of depth should just be kept 2d and crisp whilst the shots and scenes that have a greater sense of depth are really polished; this may even be how its handled I don't know the exact ins and outs as conversion is not my field (wouldn't want it to be either)

I'm working with a true stereo movie at the moment and its not that unusual for me to query whether the playback on a shot is properly stereo as I don't notice any depth . So I'll even flip between the single eye images just to check.

You can also have a significant amount of control over the depth in a shot even when its been converted ( almost more than the true stereo stuff).

The latest Potter films ( I got bored with them after working on the first 4 so now avoid working on them) would be good ones to check out. Likewise the Phantom Menace when that comes out . (If the guys at ILM are happy with the stereo conversion , I would suggest that conversion has somewhat matured sufficiently. Bay for example has been very negative in the past regarding conversion but he's now using it for a pretty large ratio of shots on Transformers 3. And thats with the same company that did Clash.

You can also get situations when the client (who is somethimes not the most discerning individual depending on background) really forces the 3d effect beyond what would be reasonable given the nature of the material.

I'm not a massive fan of older films being converted myself but I'm positive if films are shot 2d but with an eye as to how they would behave in stereo ( they composition as if the film is stereo but only use a single eye view) then the subsequently converted films would look little different to true stereo versions.

When you see the sort of angular offset on true stereo imagery and how tiny it is even on shots with a large sense of depth its not difficult to understand how relatively straight forward it is to generate 3d from 2d. And its unfair to suggest the conversion process is sloppy , I've seen the lengths that they go to to get the stuff acceptable , they have pretty high quality standards even in comparisson to other areas of film-making.
 

mike7

Distinguished Member
Thanks for a very interesting read. It re-inforces my view that interest in 3D cinema is already starting to fall off and the recent Pirates film did little to strengthen the case. The 3D special effects, which people like to see, eventually do get a bit boring and they are usually enhanced by CGI rather than straight camera work. Avatar relied heavily on this. The majority of 3D films are animations where manipulation of depth is easily created by a computer. If 3D is to have any future then we need to watch the efforts of the live broadcasts, mainly of sporting events. To a straight drama movie it is hardly worth the extra production expense and with home viewing always gives the impression, to me, of looking in a box.
 
Last edited:

Louis Mazzini

Active Member
I’m encouraged by your optimism for conversions. As you suggest, Menace should be a flagship title for the process. I can’t remember the exact timeframe, but I’m fairly sure Lucas first started talking about converting his films to 3D several years back, certainly long before the current “boom” began; I’d be curious to see some of his earliest test footage to see how it shapes up against the latest work.

That’s interesting about Transformers featuring a large amount of converted shots: Possibly the most fun I’ll have with that film now will be trying to spot those converted bits. (Incidentally, I fully agree with anyone who’d tell me that that’s no way to watch a film, but pretty much the only reason I’d go to see a film like that in the first place is to check out the 3D.)

As someone who’s been dabbling with stereoscopic photography for thirty odd years, I’ve always been drawn towards both hyperstereo pictures of large distant objects and hypostereo pictures of small close objects. I think I tend to find these more interesting than standard 65mm separation shots of medium distance objects because they offer views that we can't possibly see in the real world with our own eyes. Some recent true stereo films have had a very effective heightened sense of depth throughout (such as Resident Evil), whereas Tron seemed to have been made mostly with a very small stereo base to give it an extremely shallow sense of depth which, for me, seemed a bit pointless.

That’s a good point about conversions giving the film-maker more control over the amount of depth they want to put in during post-production, rather than being locked into a specific amount of depth that would be determined during shooting with a stereo camera. I still remain to be convinced, however, that the infinite number of extremely subtle differences between left & right eye images that can give depth and solidity to everything on screen, rather than just a sense of space between flat layers, can ever be adequately re-created from a 2D image, no matter how sophisticated the conversion process. I’ll try to keep an open mind until Menace and Titanic, which could be make of break for the process – assuming the whole 3D thing, both true stereo and converted, isn’t already broken by then through lack of interest anyway…
 

Steve Withers

Reviewer
Well it's good to see that my article has succeeded in its main purpose of generating some informed debate on the subject of 3D.

Going back to Mr.D's original point about the complexities of shooting in 3D, clearly I couldn't cover all of them in one article. The main point I was trying to make is that 3D filmmaking is still in its infancy and the purpose of the Sony course (and others like it) is to help content creators to deal with the difficulties that arise when shooting stereoscopically. I have watched quite a few live action stereoscopic films recently and as Mr D points out things like mismatched reflections can cause problems for the filmmakers. However despite the obstacles I still think it is a worthwhile endeavour to shoot films in native 3D rather than converting them in post-production.

The reason that I mentioned 2D to 3D conversion in the article was merely to draw attention to what I feel is a worrying trend at the moment. I appreciate that the 2D to 3D conversion of Clash of the Titans was rushed in six weeks at the behest of the studio so it probably isn't the best example but the truth is that I have yet to see a 2D to 3D conversion that I couldn't recognise as such within a few seconds. Accurately recreating a 3D image from a 2D source is incredibly difficult because it is the almost infinitesimal differences between each eye's view that gives everything we see a sense of reality.

I have read James Cameron's comments on the subject and he has certainly referred to conversions as '2.5D' in the past but since he's currently plugging the re-release of Titanic I'll hold judgement until I've actually seen it next year. The same goes for Peter Jackson and Michael Bay who are hardly going to criticise conversions whilst promoting their own use of the process. As for George Lucas, well he lost any credibility years ago and the best thing he can do with The Phantom Menace is burn it and start again. No amount of 3D conversion is going to polish up that particular turd.

As Bill points out the final Potter film should be a good bellwether but as I said I have yet to see a convincing 2D to 3D conversion - did you have any specific examples in mind Mr.D? As an aside I notice that whilst Piranha was shot with 3D in mind but using 2D cameras and converted in post, the sequel is actually being shot using 3D cameras.

I think that different filmmakers can use 3D in different ways and certainly the hyperstereo used in Resident Evil lent itself to that particular film but would obviously be out of place in a more naturalistic movie. I was surprised at the lack of depth in Tron, I would have thought in a film where the real world is in 2D and the Grid is in 3D you would want the 3D effects to be more exaggerated since the Grid is an artificial environment anyway. I know this is a contrarian view but I found the 3D in Tron to be a huge disappointment.

I very glad to hear that the BBC have changed their minds and will now be broadcasting the Men's and Women's Finals from Wimbledon in 3D, this will be an important event in the development of 3D as a mass market format.
 
Last edited:

Mr.D

Distinguished Member
W

As Bill points out the final Potter film should be a good bellwether but I have yet to see a convincing 2D to 3D conversion - did you have any specific examples in mind Mr.D?


Nothing beyond the forthcoming examples already mentioned. There are a lot of films I can't tell you about (back catalogue stuff) that are coming too. Very hefty blockbusters , quite a few from very eminent directors who have been very surprised by the 2d to 3d tests done on their films.

In the past I've been extremely harsh in my criticism of stereo conversion but I'm now at the stage where its getting tricky to differentiate real stereo from converted in blind tests. I'm also very gratified as to how precise and elaborate the conversion QC process is. Very subtle errors are quite rigorously corrected. And I'm not exagerating when I say its an army of people that do these conversions ...upwards of 3000 people.

Some of the other companies have been known to buy a £30k convertor box and charge clients £2million for conversion.

As I said I'm currently working on a true stereo production and the actual difference in information left to right eye are extremely small even for massively deep imagery. More often than not too much information revealed by the alternate eye causes oclusion and disparity issues (inconsistent edge depth , ghostly floating details) all requiring considrable manual tweakery.

My advice is to not give up on conversion just yet and see what the next 18 months brings.
 

Steve Withers

Reviewer
It will certainly be fun to play 'spot the converted shot' in Transformers 3 and given the decent post-production period that the final Potter film has had it will be interesting to see how that looks. However given the box office numbers for Pirates I think Potter might be a bellwether in an entirely different manner. If people continue to favour the 2D version - even for a film like Pirates that was actually shot using 3D cameras - then you have to wonder how much longer the studios will continue to support 3D. As we've discussed shooting in 3D is problematic and I'm assuming these labour intensive 2D to 3D conversions aren't cheap so if the box office isn't there then what's the point?

I'll also be watching the BBC's 3D broadcast of Wimbledon with great interest and especially the public response to it. To date the public interest in 3D at home has been indifferent at best and there is a lot of resistance to wearing glasses. I know for a fact that the sales of 3D displays have been far below the manufacturers expectations. This is undoubtedly due to a combination of apathy and the current economic environment but I can't help but feel that 2011 might be the make or break year for 3D.
 

Mr.D

Distinguished Member
2D To 3D Converted Movies - good enough?

This is an interesting article...unfortunately for all the wrong reasons some of which I've touched on.

The writer draws attention to the lack of depth in certain areas of shots and says that shooting them in true stereo would have given a better result. This is not necessarily true . For example shiny pebbly surfaces are hell to shoot in stereo because even with polarisers the disparity between the highlights on the pebbles can be large enough to turn it into an odd ghosting mess.

Often true stereo material of similar subject matter ( and shiny leather outfits ...tron legacy for example) has to have the highlight detail removed in post or transfered from one eye image into the other so the net result is that the area can end up looking flat but at least its not makeing your brain confused. Stereoscopic rigs tend to have polarisers on the front (not to provide seperation but to knock back the anglular disparity of reflections...it doesn't really work that well).

Areas that are out of focus (even slightly) tend not to have enough edge delineation to give them depth (he mentions the picture frame on the wall in Harry Potter for example). Or because of misalignment on the stereo cameras they produce floating disparities which are then killed off in post with flattening being the lesser of two evils.

The writer says that stereoscopic cameras are easy to handle....this is manifestly not true ...a single camera is tough enough to rig correctly and speedily on a film set let alone on location ( you will usually see at least 3 people operating a normal 2d camera on a film set and another 3 guys hanging back to run in with new lenses or reset the camera position. Complicate that with stereo functionality and you have a real pain to deal with that alos eats into shoot time ( the cardinal sin on a film shoot).

The writer mentions things like Nuke having sterescopic capabilities , which it does (along with other packages its not anything unusual for the last 5 years) however its just capable of displaying the imagery and managing the stereo workflow its not really automating any of the work in terms of the additional stereo manipulation. The systems that do claim to do this are not really of much use without lots of manual intervention anyway.

Also in my experience the actual integration of the alternate eye image ( usually work is signed off on a single eye and then the stereo transfer proceeds) is somewhat straight forward if you were working efficiently when you commenced the work. Its not working with stereo thats a problem its the nature of the stereo imagery itself.

So the assertion that shooting stereo is a panacea for some of the problems that people ascribe to conversions is not as cut and dried as people make out.

At the moment I'd say that conversion and true stereo are both as problematic as the other (as witnessed by people coming away from true stereo films with experiences of lack of depth or strange occlusion type artifacts.) However there is not a lot of evolution likley to happen in true stereo rigs whereas the conversion process is being improved constantly in line with processing power and more compatable ways of shooting 2d for conversion becoming more widely employed.

I forsee a day when the camera on set is really just capturing telemetry and scene reference ( and performance reference) with the end result on screen being an entirely rendered 3d model using the initial captured data as reference. Avatar pretty much did this at least in terms of the main characters. So you no longer see the real world but the real world as rebuilt by a computer .

This may sound anathema to we who are used to cinematic depiction through the camera lens. Is it really that different to light being turned into electrical signals which are then quantified by digital bits which are then further manipulated extensively in post compared with it being a chemical reaction to light on a very specially constructed and evolved chemical sandwich?

The whole idea that what you see through a camera lens is somehow close to reality is somewhat facile, if you ever stand on a film set you will readily appreciate the synthetic nature of film-making and that the resultant fabricated depiction of reality is no closer to representing your actual experience of the real world than the ink and paper of a book that can also put imagery into your head.

Eventually the notional intrusion of "computer vision" will diminish and we will accept imagery translated through a silicon engine in much the same way that we accept a mechanical , electrical or chemical imaging engine today. At that point 3d depiction becomes trivial and can we really say that the data captured from the real world is less relevant or accurate than the data collected in those other techniques we are familiar with and even cherish today?
 

phil5366

Banned
Wow! What a great topic. Some very good informative writing there chaps and, lets have some more.
 

Steve Withers

Reviewer
I caught the BBC testing side-by-side 3D on their dedicated HD channel today and it looked pretty good, I'm looking forward to watching the Wimbledon finals in 3D.
 

Mr.D

Distinguished Member
I caught the BBC testing side-by-side 3D on their dedicated HD channel today and it looked pretty good, I'm looking forward to watching the Wimbledon finals in 3D.

I'm having to make SBS from RED footage to get a panasonic VT20 to function as a preview monitor for a feature film. The playback software does frame sequential just fine , unfortunately the plasma won't give me any mode that will keep the glasses active without a correct hdmi flag and not mess the picture so I had to fake up SBS.

This is admittedly an improvement in status from the previous "expert" who delared the Panasonic plasma to have "broken 3d" whilst squinting at it though a pair of passive polarised glasses with "JVC" written on them :rolleyes:

The plasma is okay but to really check edges and things I need to use a 2k FS christie projector in a DI suite. If I could get the plasma to make use of the full res it would speed up my workflow a fair old bit as I can only get into the DI suite twice a day.

This is my first stereo film and I have to say the actual mystery of "stereography" is vastly exagerated by the so called experts. Pretty much had the issues understood and even some better techniques developed in 48 hours (just by examining and thinking about it myself)
 

Steve Withers

Reviewer
I'm not sure I understand the point of that post, other than to rather patronisingly contradict your earlier post about how difficult it is to shoot in 3D?
 

Mr.D

Distinguished Member
I'm not sure I understand the point of that post, other than to rather patronisingly contradict your earlier post about how difficult it is to shoot in 3D?


I made no comment in that post as to the practicalities , difficult or not of shooting stereo. I've referred to "stereography" to encompass the whole range of fundamentals required to deal with stereo footage effectively. Its obvious from my post I'm dealing with footage that's already been shot.

My point was to highlight the limitations of SBS more than anything.

As to why you may find this patronising I have really no idea.
 

Treguard

Active Member
What a great topic :) I just hope though that companies don't forget those of us with strabismus who can't see 3d to the effect you guys can :)
 

Steve Withers

Reviewer
This will be an interesting weekend for 3D with the release of Transformers 3 and the BBC broadcasting the Wimbledon finals in 3D.
 

Steve Withers

Reviewer
Well the Wimbledon broadcast was something of a mixed bag, it was an interesting experiment but whilst the low camera angles made it look exciting it was also difficult to tell what was happening on the other side of the court. A few more camera angles would have been nice and the BBC also used 2D to 3D conversion a few times which was very obvious. Another problem was that since the players were wearing white it made crosstalk more obvious but overall the 3D looked very good in the broadcast. Still it was interesting to see the Sony 3D cameras in action and to see the BBC using many of the techniques discussed on the Sony 3D course.
 

Steve Withers

Reviewer
I went to see Transformers 3D last night in the hope of seeing some good 3D because to be honest I hated the first two Transformers movies. Unfortunately not only was the film just as bad as the other two with non-sensical plot, boring fights between robots that I couldn't tell apart and characters I couldn't care less about but worse than that the only good thing about the first two films (Megan Fox) had been fired and replaced by an English Victoria Secrets model who couldn't act her way out of a wet paper bag. To cap off a terrible experience the 3D was awful as well, with backgrounds that were often inverted, appalling colours, far too much camera movement and fast editing and a lot of very bad 2D to 3D conversion. In fact most of the 3D looked like a conversion with very little sense of depth, eventually my friend and I couldn't take anymore and about half way through we just walked out.
 
Last edited:

The latest video from AVForums

Podcast: Panasonic JZ2000 Final Thoughts - TV Calibration: Should you? And More...
Subscribe to our YouTube channel

Latest News

What's new on Blu-ray Special - The Cinema of Zhang Yimou & Gong Li
  • By Mark Costello
  • Published
Black Friday 2021: What you need to know
  • By Andy Bassett
  • Published
AVForums Podcast: 24th November 2021
  • By Phil Hinton
  • Published
Panasonic 2021 TV update brings full 4K resolution with VRR and HFR
  • By Andy Bassett
  • Published
Best Hi-Fi Products of 2021 - Editor's Choice Awards
  • By Ed Selley
  • Published
Support AVForums with Patreon

Top Bottom