English version White Paper 3-D in Cinema and Television v2

V2

White Paper
3-D for Cinema
and Television
TelevisiON
Document created under the sponsorship of
Document created under the sponsorship of
Ficam, the CST, UP3D, the HD-Forum and the AFC
and the AFC
Editorial Coordination ::Marc Bourhis (Ficam) and Olivier Amato (Smartjog)
Editorial Coordination Marc Bourhis (Ficam)
© 2012 Ficam

Partners for this white paper :

The Ficam (Fédération des Industries du Cinéma, de l’Audiovisuel
et du Multimédia), presided over by Thierry de Segonzac, is a
professional organization bringing together 180 enterprises whose
activity covers all the professions and technical know-how dealing
with image and sound. Always proposing strong measures, the Ficam
represents, promotes and defends the national and international
interests of technical industries dealing with audiovisual creation.

THE 3D CONSORTIUM
The Commission Supérieure Technique de l’Image et du Son (CST),
presided over by Pierre-William Glenn is an association grouping
professionals in cinema and the audiovisual and multimedia fields, launches
3D rendez -vous
as well as technicians and technical artists. The CST brings together
today close to 700 members. Its goals are to defend the quality
of the production and distribution of images and sound, defend
creativity and technological and artistic innovation in cinema and
the audiovisual fields, as well as to defend the independence and
freedom of action and expression in our professional activities.

A UNIQUE MARKET PLACE
UP3D (Union des professionnels de la 3Ds) is an association whose
objectives are: to bring together and animate the 3-D professional
ONE TO ONE MEETING - BUSINESS NETWORKING
community, inform and communicate on the professions, know-
how and equipment specific to the 3-D world, and to promote and WORKSHOPS - BOARD DEMO PRESENTATION
develop 3-D in Europe.

Originally composed of 17 members, the HD Forum, presided over
6 ZONES
by Jean-Pierre Lacotte, now has 51 members, all professionals
as stated in the statutes. The diversity of the questions to resolve
around HD has led the HD Forum to create two commissions: the
MIDDLE EAST - BRASIL - NORTH AMERICA
“technical commission” and the “communication and marketing
commission”, which are themselves divided into work groups that
RUSSIA - INDIA - CHINA
regularly collaborate with the work of the Ficam and the CST.
FOUNDED BY :

The AFC (Association Française des Directeurs de la Photographie
Cinématographique), co-presided over by Matthieu Poirot-Delpech,
Michel Abramowicz and Remy Chevrin, brings together most of the SPONSORED BY :
directors of French cinematic photography at the highest artistic
and technical level of French and foreign productions. The AFC
promotes the existence of quality cinematic images, tests new
techniques and affirms the competence of photography directors
as creative collaborators of film directors, in the best tradition of
French cultural and artistic discourse.
contact@3dconsortium.biz

Special thanks :
To Pascal Buron (associate president of Ficam in charge of
the Research and Innovation technical commission, and to Olivier
Amato (Smartjog) for all the coordination work they did on this
project as they have done on so many others.
A big thank you to Yves Pupulin (Bincole) for his contribution

TOMORROW
to all the chapters of this white paper. Thank you also to Olivier
Cahen (Stéréo-Club Français) for having shared with us his
encyclopedic knowledge of 3-D and for his editorial contribution to

starts here.
the visual comfort chapter, as well as to Philippe Gérard (3DLized),
who, through his commitment to research on 3-D visual comfort,
permitted us to be at the forefront of ideas on this question.
A big thank you, too, to Marc Léger (INA Sup) for the writing
of the chapter on the main principles of 3-D and the updating of the
chapter on codecs and modes of TV distribution.
Many thanks to the late Alain Derobe (stereographer and
photography director) who passed away on 11 March 2012, for having
accepted in due course the taking up again of his contribution
made to the CST concerning the adapting of 3-D to the size of the
broadcast screen, as well as to the members of the CST for their
contribution to the part on 3-D projection for cinema.
Thanks as well to Laurent Verduci (stereographer) for his
educational aids on 3-D which we used as a common theme and ©2012 Cisco Systems, Inc. All Rights Reserved

for his active participation in our work groups. Thanks also to Alain
Today, it s easy to marvel at how ar we ve come.
Chaptal (Sonovision), Franck Montagné (Imagie), Michel Chabrol Our phones talk to our TVs to record our avorite shows. Doctors in Estonia diagnose
(Eutelsat), Thierry Gruszka (NDS/CISCO), Eric Martin (Technicolor), patients in Denmark. Social networks help companies improve customer service.
Charles de Cayeux (France Télévisions), Pascal Charpentier And yet, up to now, more than 99% o our world is not connected to the Internet.

(UP3D) and Benoit Michel (stereoscopynews.com) for their active But we re working on it.
And tomorrow, we ll wake up pretty much everything else you can imagine.
participation.
Trees will talk to networks will talk to scientists about climate change.
Finally, we wish to thank all those who attentively reread
Stoplights will talk to cars will talk to road sensors about increasing tra ic e iciency.
the final document. Thanks also to David Steiner and Cédric- Ambulances will talk to patient records will talk to doctors about saving lives.
Alexandre Saudinos (Parallell Cinéma) for their contribution on the It s a phenomenon we call the Internet o Everything an unprecedented opportunity
techniques of preproduction and filming, as well as to Philippe Ros or today s businesses.
(photo AFC director).. Tomorrow?
We re going to wake the world up. And watch, with eyes wide, as it gets to work.
#tomorrowstartshere

1

SOMMAIRE 0

0
FÉDÉRER ET PROMOUVOIR LES INDUSTRIES TECHNIQUES
DU CINÉMA, DE L’AUDIOVISUEL ET DU MULTIMÉDIA
1
Why a White Paper on Stereoscopic 3-D* (3Ds)?? PAGE 8 1

ma

0
Technical industries iné
Fundamental Principles of Stereoscopy PAGE 10

0
c
of creation audiovisuel
RIGS AND STEREOSCOPIC CAMERAS PAGE 20 1

LIVE MULTI-CAMERA 3DS CAPTURING PAGE 24

0
00
1

3Ds POSTPRODUCTION: CORRECTIONS AND WORKFLOW 1
PAGE 25
00
1
1
1

2-D/3Ds CONVERSION: THE NEED FOR KNOW-HOW PAGE 30
1

0
0
0

VISUAL COMFORT WITH 3Ds PAGE 32
0
1 00
1 0 1
0

3Ds PROJECTION FOR FILMS PAGE 34
mult
imédi
3-D TV DISPLAY PAGE 36 a

0
ADAPTING 3-D TO THE SIZE OF THE BROADCAST SCREEN PAGE 40

3Ds GLOSSARY PAGE 45
0 With over 180 member companies,
APPENDIX 1: DEFINITION OF 3Ds PROFESSIONS PAGE 47
the Ficam represents and promotes
the technical industries of Cinema,
APPENDIX 2: 3-D FILMING TECHNIQUES PAGE 49 Audiovisual and Multimedia.

APPENDIX 3: CODECS AND DISTRIBUTION MODES PAGE 54
www.ficam.fr

FICAM (Fédération des Industries du Cinéma, de l’Audiovisuel et du Multimédia)
11/17, rue de l'Amiral Hamelin-75783 Paris Cedex 16-Tel : + 33 (0)1 45 05 72 55-Fax : + 33 (0)1 45 05 72 50

8 9

Why a White Paper on Stereoscopic 3-D* (3Ds) ?

E
ver since movie theaters have been equipped with digital 3-D pro- production. In fact, stereoscopic 3-D, if it adds new possibilities on the
jectors, a number of 3-D feature-length films have been shown narrative level, also includes new obligations on the technical level tied
with often disparate processing quality. The television market has to the cerebral acceptance of cinema and television viewers. This is why,
also decided to respond rapidly to the desire of the public for 3-D films. in this 3Ds white paper, our objective is not to constrain the creativity
The television manufacturers, quick to participate in this technological of directors, who should remain free in their artistic content, but to set
craze, now make flat television screens on a grand scale equipped with a framework for their being better able to get around in the additional
display systems “in 3-D”, allowing the viewer to watch 3Ds TV programs production possibilities specific to stereoscopic 3-D.
in conditions of quite acceptable visual comfort. Moreover, some smart- It is also important to specify that the study of stereoscopy,
phones have recently appeared in the mass market which are capable already quite old, has allowed the visual comfort zones–inside which a
of generating autostereoscopic 3-D. viewer will have no discomfort or, not to say headaches–to be empiri-
However, the market for mobile, TV and cinema broadcast cally defined. Scientific studies are now taking place to refine these pa-
hardware is much in advance, chronologically speaking, of the produc- rameters.
tion capacity of stereoscopic 3-D content by the audiovisual and cinema We say also that the art of 3Ds is only at its beginnings. There will no
industry. Up to now, there does not exist enough natively produced 3Ds doubt be the possibility for directors of films or television broadcasts to
content to furnish all the ad hoc broadcast media. The barrier is mainly go beyond this technical framework, at least from time to time, in order
economic and technological, but it is also a result of a lack of informa- to provoke quite new emotions in the viewer.
tion on the practical conditions in which it is possible today to create Finally, it should be noted that the perception of the depth of an
film in 3-D for television and cinema. object in a 3Ds scene is modified according to the size of the display of
Due to the current deficiency in 3Ds content, it should be noted the images (movie theater screens or televisions). This white paper thus
that many 3-D films and programs broadcast in movie theaters, or on describes the different display technologies and defines the technical
the first 3Ds television channels, use procedures that create 3-D from contours of future 3Ds programs to be provided for television and the
cinema.
2-D automated images. These do not sufficiently take into account the
In fact, in order to summarize the main objective of this white
main principles of visual comfort for the viewer and put into question
paper, we would say that it consists of sufficiently informing audiovi-
the notion of respect for the original work.
sual and cinema professionals on the possibilities of 3Ds so that they
Given this enthusiasm for 3-D, it seemed necessary for diffe-
can appropriate this new language as much as possible, and that public
rent professional organizations of the cinematographic and television
disappointment can be avoided for this new artistic and technical field,
industries (Ficam, CST, UP3D, AFC, HD-Forum, etc.) to produce a white
as might have been the case during the first wave of infatuation for 3-D
paper laying out the current state of stereoscopic 3-D. The goal of this films in the 1950’s.
white paper is to point out the different technical and narrative issues
tied to the filming, postproduction, visual comfort, 2-D/3Ds conversion This white paper on 3-D refers to the state of the art in May 2012. It is the
and display on a cinema and/or television screen. second version of the document that takes into account the remarks of
Stereoscopic 3-D is an artistic and technical mechanism requi- French professionals and updates the generic ideas on the subject.
ring a mental reconstruction of images, which is chosen by the director
and limited by the physiological laws specific to most viewers.
As such, it is crucial to put in place, at every step in the manu- * Rather than use the terms “relief” or “3-D”, quite imprecise for the
facturing chain, 3-D monitoring and quality procedures supervised by former and source of confusion with computer generated images for the
qualified personnel and trained in this new mode of creation and post- latter, we recommend using the term “3Ds” for stereoscopic 3-D.

10 11

Fundamental Principles of
Stereoscopy
How do we perceive 3-D ?

L
et’s look at a close object: for example Now, while continuing to look at the tip of our
the index finger of our hand. Naturally, finger, let us consider the objects situated
the optic axes of our eyes convergent behind our finger (Figure 2). On which side
towards the object being looked at. We thus to they seem to be placed when we close one
become slightly cross-eyed. The more the ob- eye and then the other? We notice that they
ject is close to us, the more we become cross- are more to the left seen from the left eye and
eyed: the angle of convergence increases. The more to the right seem from the right eye.
more the object moves away, the more the It is what is called parallax. It is said that for
angle of convergence decreases. The angle objects situated behind the plane of conver-
of convergence is a very important indication gence, the parallax is positive. Let us observe
for our brain: knowing the distance between again our finger and now consider the objects
figure 1 figure 2 figure 3
the two eyes, that are closer
it can deduce (Figure 3). The
the distance of phenomenon
the object, like is inversed: the
a range-finder. object is more to
The brain allows the right seem
us to perceive from the left
the depth of the eye and more
scene by way to the left seem
of the differen- from the right
ces between eye. For objects
the angles of situated in front
convergence of of the plane of
the different ob- convergence, the
jects being observed. Now let us examine the parallax is negative. In sum, the parallax is
images formed by our two eyes: as the optic zero for all the objects situated in the plane
axes of our eyes converge towards our finger, of convergence, positive for all the objects si-
the image of our finger is formed at the same tuated behind the plane of convergence and
place on our two retinas (Figure 1). We call the negative for all the objects situated in front of figure 4
same points of the objects in the left and right the plane of convergence.
are then practically parallel and the angle of object can hardly be perceived. Happily, other
images “homologous points”. It is said in this The more an object is distant, the more the
convergence becomes too small to inform us indications inform us of the depth: occultation,
case that the homologous points are fused perception of its distance is imprecise. When
sufficiently as to the distance of the object. Be- relative size of the objects, perspective, the ef-
and that our finger is in the “plane of conver- the distance of the object goes beyond several
ginning from a 100 meters, the distance of an fect of mist, etc.
gence” of our eyes. dozen meters, the two optic axes of our eyes

12 13

consequently the placing of the plane of
How is 3-D reconstructed? 3Ds applied to cinema Stereography : a new production language convergence as well as the amplitude of
the “scenic box” (see Appendix 2).

W
and television Creating for the cinema or for television is
It should be noted that the size of the 3Ds

T
e now take two cameras in pla- not tied only to the techniques of produc-
he fundamental difference between tion or broadcasting, but also to the emo- monitoring screen during filming does not
ce of our two eyes along with a attempts in the past and the present necessarily correspond to that of the fi-
tional connection that these images foster
3-D screen system whose role growth of 3Ds comes from the tech- nal broadcast. The stereographer should
in the human brain. From this standpoint,
is to make each eye see only the image nical mastery of 3-D in the entire chain of 3Ds requires that this connection be re- clearly inform the production team of the
that is intended for it. In Figure 4, the left the manufacturing and image broadcasting considered so as to adapt the conditions differences in the sense of depth according
and right homologous points are repre- process. The use of computer tools during of creating a cinematographic or televi- to the dimensions of the film’s intended
sented respectively in red and cyan. If the shooting and the possibilities of modifying sion project to this question. 3Ds induces screen and to justify his “not visual” but
homologous points are superimposed du- images pixel by pixel up to broadcasting a rupture with the practices of production “provisional” adjustments.
ring projection, the view will converge on are steps that from now on are indispensa- and which modifies the interpretation of The parameters of “depth” can be in-
ble and in constant evolution. They guaran- the space as imagined by the director. One creasingly modified in postproduction. Gi-
the screen and the object will be seen on
tee the quality of the 3Ds image necessary should also note that, in studying certain ven this fact, just as the color calibration
the plane of the screen. That is how one for the visual comfort of the viewer. gives the tone of the scenes in a film, the
filming parameters such as lighting and
sees films in 2-D. Digital tools take part in the mastery of “3Ds calibration” becomes an important
framing, if there is a rupture between 2-D
If the parallax is positive, the view will stereoscopic production. Being more and 3Ds, it is situated more on the side of step after the film has been edited. It al-
converge towards a point situated behind knowledgeable from this experience, we acceptance by the viewer’s brain than that lows for the perception of reality in 3-D to
the screen: the viewer will thus have the wish today to support the emergence of a of artistic practice. be given a general tonality and to provide
impression that the object is situated new cinematographic and television lan- In 2-D as in 3Ds, the director remains the 3Ds continuity between shots according to
behind the screen (as if seen through a guage tied to the usage of 3Ds. main judge of the use of 3-D in order to the final intention of the director.
window). Finally, if the parallax is nega- A purely technical conception of 3Ds is serve his narrative, whether it is a question
a risky bet. It is preferable to rely on the of a film, a telefilm or a televised program. Think about 3Ds before filming
tive, the view will converge towards a point
savoir-faire, both technical and artistic, of In the same way that he decides a produc- If you want to have a quality 3-D effect, it
situated in front of the screen, and the the stereographers and technical service tion in 2-D, he formulates a particular re- would be useful to verify beforehand the
object will thus seem to be in front of the companies, which have both a solid theo- quest to the stereographer who will see to pertinence of the choice of 3Ds for this or
screen (as if popping out). retical knowledge and experience in 3Ds. its completion. The working together of the that film or television program from the
In sum: The production of a sequence in 3Ds can- director and the stereographer thus aims beginning of the synopsis or scenario. In
• Objects situated on the plane of conver- not be summed up in an automatic algo- to establish “3-D behavior” (modulation of this way, a story unfolding in an aerial or
gence during filming will be reconstructed rithm for creating 3-D effects, however the stereoscopic range), done through a aquatic universe lends itself totally to a
on the plane of the screen (parallax zero). sophisticated it may be, or do without the graphic image or annotations on the script popping-out effect, for the objects can be
• Objects situated behind the plane of onsite experience of specialized professio- or storyboard, which then serves as a guide “unglued” from any support, like levita-
nals who know how to adapt to particular for the team doing the shooting. 3-D that ting in a fluid or a vacuum. Furthermore,
convergence during filming will be re-
filming conditions, very often full of unfo- is too strong can lead to visual discomfort longer shots and/or rather slow and fluid
constructed behind the screen: window reseen events. The creation of a 3Ds scene over time, while 3-D that is too weak risks movements of the cameras also favor the
effect (positive parallax). requires a perfectly regulated production frustrating the viewer. It is mainly within perception of a more comfortable 3-D ef-
• Objects situated in front of the plane of with the camera angles completely maste- these limits that the director can choose fect. They leave time for the public to dis-
convergence during filming will be recons- red. If the production teams are not sup- the depths he wants. The streographer cover and enter into the depth of the ima-
tructed in front of the screen: popping-out ported by competent people in this way, translates these choices using the tools ge. In the same way, wide compositions
effect (negative parallax). they risk having their possibilities limited at his disposal, tools which he will super- with framing using high-angle or low-an-
as to the creation of emotions in the viewer vise in their practical use. Given this, he gle shots favor 3-D effects and intensify
specifically tied to 3-D effects of depth or is responsible for the adjustment of the sensations of dizziness. Geometric forms,
of popping out, or even in certain cases, to angulation and the inter-axial distance buildings for example, are more quic-
provoke visual discomfort in the public. between the center of the camera lenses kly identified by our brain and contribute
according to the chosen focal length, and to the quick immersion into 3-D. As for

14 15

Creating a 3-D Scene streaming programs, among the first ex- ted cases:
periments taking place today using 3Ds in - The shot itself, as in a classic film. Ideal-
retransmission of sporting events, football ly, the storyboard creator should create a
From the beginning of writing the scenario, 3-D can be taken into account and used as a is looked on as the favorite due to the high storyboard that is optically correct.
narrative tool. It is possible to imagine scenes whose dramatic intensity will be amplified if number of viewers and a public potentially - A view “from the side”. Starting from the
one plays with the depth, the popping-out effect and the effects of gigantic size and minia- interested by the fact of viewing matches principle that the 3Ds window is projec-
turization. The scriptwriters can think about 3-D effects in advance, sometimes integrating in 3-D. Nevertheless, the current produc- ted onto the plane of the screen, this side
their ideas directly into the scenario. These can be indicated in capital letters (in the Anglo- tions of this sport on television, centered view will mark the location of the plane of
Saxon world, they are sometimes preceded by the term “3DFX”). Nevertheless, the logical on lateral panoramas and a depth of field the screen so as to distinguish the objects
process of creating a film lends to entrusting the control of stereoscopic effects to the director often difficult to master, is not necessarily that should be in front of the plane of the
and/or stereographer. For example, in Océanosaures (Sea Rex), a 3-D IMAX© film written the best adapted to retransmission in 3D, screen (popping out) and those that should
and directed by the Frenchmen Pascal Vuong and Ronan Chapalain, the stereoscopic filming unless the camera angles and the pano- be behind it (in depth).
proposals appear right from the storyboard. ramas are comple- - In the most
tely rethought out complex cases, a
OcéanoSaures - Voyage au Temps des Dinosaures © 2010 - N3D LAND Productions

with respect to view from above,
It should be noted in the case of this film a 2-D retrans- with the location
that Pascal Vuong created the storyboard mission. A fur- of the plane of the
himself and Ronan Chapalain supervised ther constraint screen, could also
the stereography, thus allowing each shot comes from the be useful.
to be conceived very early in all its depth fact that the ac- Beyond these
and not as a 2-D image. In this storyboard tion takes place general rules,

© Convergence3D.net
extract, the colored markings added to the most of the time it could also be
initial drawing permit one to quickly loca- far from the ca- advisable to use
te the elements to be popping out and the meras, thus re- previsualization
position of individuals in relation to the quiring the use software that si-
plane of the screen. This stereoscopic pro- of a long focal mulates what the
duction information, added to the framing length on the basis shot or perspec-
and cropping choices, allows the director of wide 3-D shots Screen capture of 3Ds previsualization software tive will be in or-
and film crew to better prepare the shoo- that could create a der to define the
ting and postproduction of the images and “maquette” effect. camera angles
prevent certain pitfalls of stereography en- On the other hand, certain inside sports, and movements in 3Ds. Such software al-
visaged too late in the process. such as basketball, where one can be clo- lows computer-generated images to preci-
ser to the action, offers 3-D that reinforces sely reconstitute each filming location and
the viewer’s emotions. the camera position (whatever the type of
rig). It allows the frame, the inter-axial dis-
Preproduction tance, and the angles to be calculated and
Preparation is essential in order for fil- previsualized during filming and in post-
ming in 3Ds to be successful. Researching production. All the parameters, such as
particularly meticulous points (measuring the focal length, the diaphragm, the depth
distances, verification of distant objects, of field, the movements, etc., can also be
taking into account the size of the 3Ds rig) predetermined.
permits many problems to be avoided in
advance. Creating a 3Ds storyboard is ad- The importance of checking volumes
vised. Without help from software, this sto- During the work of identifying certain para-
ryboard could show, for each shot, at least meters and of preparing a film in 3Ds, it is
two or more views for the most complica- necessary for the director and the head of

16 17

photography to check the frames by taking lengths, the establishing of the position of one of the fundamental parameters. It is when there is a blurred background with
into account the development and distri- the massive forms of the film set becomes what happens within the shot or out of it backlit foliage.
bution of volumes relating to depth. They even more decisive in determining the po- that justifies the movement of the camera
rely on the stereographer who helps them sition of the actors and the viewpoint of according to the director’s decision. In
in looking for homogeneity in the choice the camera in terms of the depth and the
Using visual indications of depth for 3Ds
streaming programs, the frame is deter-
of angles and focal length in relation to edge of the frame. The sensation of 3-D, productions
mined much more by following the action,
3Ds. This step should allow the position even if it appears realistic or admissible Creating a shot in 3-D does not only mean
which in sports for example, is generally
of the objects to be determined within the for a single shot, might no longer be so adjusting the stereoscopy, but also to use
centered.
“scenic box” containing the volumes of the during the editing of several shots one af- information that is going to ease the un-
scene in three dimensions. This checking ter another. For example, during filming in derstanding of the depth of the image.
is of major im- 2-D, it is normal to
A camera in movement It is thus necessary to know how to use
Moving shots reinforce the perception of both stereoscopic and non-stereoscopic
portance. “cheat” in the po-
the succession of volumes within the sce- information.
At this step, a sitioning of the
ne. This emphasis of volumes is obtained, Painting and photography remind us that
warning is neces- actors (to make a
of course, by the movement of the actors in 2-D numerous factors allow the brain
sary to point out shot or perspective
in relation to the objects in the frame. It is to analyze and understand the depth of an
that, as opposed easier). This prac-
equally increased by the camera’s view- image. This faculty is conserved, moreo-
to 2-D images, tice should arouse
point, which reinforces the presence of the ver, when one eye is closed. The utiliza-
stereoscopic the greatest vigi-
actors. Perhaps this could be a new type of tion of this non-stereoscopic information
viewing provides lance when using
space between cinema and theater? makes the understanding of space, and
the viewer with 3Ds.
the perception the participation of the sensation of 3-D,
of the position Framing Lighting in 3Ds easier to experience. Among these fac-
in depth of each If, in 2-D cinematography, light and its tors, we can cite the following:
The precision of
object, and thus contrast are only limited by the projection -An object up close masks an object far
the cinematogra-
their sizes. The standards and, in this context, the power away. This effect, which can be obtained by
phic frame is one
choice therefore of the projector, it is quite otherwise in the movement of the actors or the objects
of the essential
of the 3Ds fil- stereoscopy. Strong contrasts, in particu- in the depth of the scene, is reinforced by
parameters of 2-D
ming parameters lar at the edges of the shot, through fo- the movement of the camera’s viewpoint.
filming. Yet, with
can lead to a si- liage against the sunlight or between dif- The dollies, the movements of the cra-
stereoscopy, whe-
tuation in which ferent parts of the subject, can lead to real nes or steadicams, or the movements of
re we are nearer
a person can be cerebral discomfort for the viewer. a shoulder-held camera, emphasize the
to low and/or high
perceived as a The lowering of contrast is often advised, relative position of objects in relation to
3-D effects, the
giant (magnifi- without us knowing how to precisely each other.
frame shows a
cation) or like a Visualization of the benchmark visual comfort slight uncertainty quantify it today. This being so, shots for -The size of objects decreases with distan-
drawing on card- zone as regards 3Ds. which the lighting and contrast levels are ce. This can be used to “cheat” in certain
in relation to the
board (flattening), not controllable require sensors covering shots, provided that a likeness is preser-
two shots being
or like its opposite a large range in order to obtain images ved with the effects of reduction in size, if
filmed, which only overlap completely on
(a maquette or stretching effect). capable of conserving 3-D in high and low a realistic image is to be conserved.
the plane of convergence or the plane of
Finally, it is useful to note that, as a ge- the screen. Furthermore, the notion of the lighting. This also allows a better adjust- -The gradation of textures: bricks, gras-
neral rule, 3-D is best appreciated when immersion of the viewer for certain ste- ment of contrast during the calibration ses, pavements, etc., become less defined
the scene is in movement, close to the ob- reoscopic productions is often emphasi- of stereoscopic films and TV programs. with distance.
ject and with a rather short focal length. zed even before the notion of framing, in Likewise, it is advised to pay attention to -The “atmospheric dispersion” (mist ef-
Nevertheless, it is difficult to produce a IMAX© for example. But for most cinema- clipping of strong lighting, for such lighting fect): it is due to the pollution of the mi-
whole film without close-up shots. For tographic productions, the frame remains can produce a disparity of form between lieu being passed through according to its
shots resorting to medium and long focal the two stereoscopic images, such as thickness. It works for infinite distances

18 19

Managing the Popping-out Effect

Popping out is one of the effects is to be weak and slow if the the capacity of the viewer to
belonging to 3-D. It gives the intensity is to be strong. The “read” comfortably the effects
feeling that the objects pop same popping out can vary in of popping out: at the beginning
out of the screen towards the intensity and be produced twice: of a film, around three seconds
viewer. first weak then strong, in order are necessary for the brain to
However, there are rules to to reinforce the final effect. The submit to the interval between
follow so that this effect does not total length of the effect cannot convergence and accommodation
cause any visual discomfort. The be too long. For reasons of visual that is imposed on it, whereas
elements or parts of elements comfort and of concern for a after ten minutes of a film in 3-D
that pop out must be contained narrative balance, the frequency stereoscopy, this accommodation
within the field of the cameras. of popping out throughout a time is reduced to half a second.
The speed of the appearance of film or a TV program must be It is thus strongly advised against
popping out must not be too slow an intelligent dose between the abusing the effects of brutal
or too fast. A sufficient time must repetition of popping out and popping out at the beginning
be left for the viewer’s brain to their differences in intensity. of a film, for they can become
get accustomed to this effect. Finally, it has been noticed painful and wrongly interpreted
To preserve the visual comfort, empirically that there is a by viewers’ brains.. Modulation example of the stereoscopic range for the length of a film
it must be fast if the intensity phenomenon of learning in
Stereoscopic range and modulation of stereosco- is, moreover, variable according to the
pic ranges individual.
as well as for a smoke-filled room. understanding depth. This is why certain The stereoscopic range is a measure of Some scientific studies are, however,
-The effect of perspective is very important head operators prefer using light that is the quantity of 3-D in an image. It is the still going on in France (3-D Acceptance
information: as we look, we search in per- more directional than in 2-D in order to difference between the interval of the most & Comfort) and in the United States
manence for a scene’s convergence lines emphasize the shadows. Regions without distant object and the interval of the closest (BanksLab, University of California at
to understand its perspective by using detail of an image—for example a comple- object in a shot. Berkeley) to clear up the relation between
roads, railways and buildings. tely white sky or an interior scene with to- The stereoscopic range can vary within a stereoscopic range and visual comfort.
-3-D is above all comparative. The brain tally black zones—are less favorable to the shot, between two shots and during the Without 100% reliable reference points in
performs better in comparing, for exam- comprehension of 3-D than a cloudy sky or length of a film. Within all these cases, one this domain, we recommend, during the
ple, colors than in analyzing them. It’s the an interior in which half-light allows the can speak of “modulation of stereoscopic production of a teaser or a test, to examine,
same thing for stereoscopy. A distribution subtlety of the details to be seen. ranges”. case by case, the contentious shots that
of volumes in the depth of a scene will fa- -One must also attend to the consistency of The stereoscopic range (histogram of deviate from the “acceptable” limits of
vor the impression of viewing in 3-D. It will the screen depths and avoid the “collision” disparities) is a quantifiable data for each stereoscopy.
be more difficult if there is only one ele- of a title placed on the plane of the screen image in percentage of its size or in pixels. In general, it is customary to moderate the
ment in the foreground separated from with an object that is popping out. As for The term “depth budget” is sometimes exceeding of these limits in favor of the
another element placed at infinity without a “window violation”, it is characterized by used to designate either a stereo range or production.
any intermediary volume. For example: a the collision of the edge of the screen with the modulation of stereo ranges, which can It should be noted that there already exists
bee popping out on a background of infi- an object that is popping out. lead to confusion. technical recommendations defining the
nitely distant mountains will not favor the The same stereo range gives a weak acceptable limits of stereoscopic ranges
impression of viewing in 3-D. It is important to use all of this information to impression of 3-D on a small screen, for 3Ds TV broadcasting (ex: BSkyB
-Furthermore, it is sometimes preferable 3-D). The establishment of technical
enrich the stereoscopic production parameters. stronger on a movie screen and very
to have titles popping out on a textured powerful on a giant screen such as recommendations is currently under study
background rather than on a black one. In a general way, the least contradiction between
IMAX©. The acceptance by the viewer of in France.
-Textures and shadows are stereoscopic monocular and binocular (that is, parallax) depth
a stereo range chosen for a given screen
objects, supplementary “information” for information cannot be allowed.

20 21

Ensuring the continuity of the modulation of the each other on more or less sophisticated
stereoscopic range during a whole film mechanical or motorized rigs. The advantage
It is indispensable to think of a 3Ds film or of this arrangement is that it is often more
TV program in its entire continuity. For ci- rigid and avoids the aberrations linked to the
nema in 3Ds, like in 2-D, the order of se- mirror set up.
quences and shots at different steps in the Its inconvenience is that it imposes wide inter-
making of a film are often modified, and axial distances which results in a “maquette”

Binocle
particularly when it is being edited. effect tied to the apparent miniaturization of
The modulation of the stereoscopic range the objects being filmed. It is not known how
may be modified by the inversion of shots to avoid this effect today, but it comes into
during editing. For example, a close-up play from the moment one goes beyond the Two types of side-by-side rigs
with much popping out will correspond ba- inter-axial distance of human vision. It can be
dly with a wide landscape shot having a lot put up with, or even looked for, according to
of depth: this alternation tires the viewer. how the director wants his production to be. shots in a fiction film. advantage of having its vertical camera placed
It is thus necessary to take care in the or- In any case, there is no other solution when Mirror rigs are used with steadicams, shoul- above the mirror, allowing the optical axis to
ganization of the shots to ensure smooth one films from far away, with objects in the be brought very close to the ground for ma-
der-held cameras and most classic machi-
transitions of depth for the public. foreground situated far from the camera king large vertical low-angle shots.
nery. Mirror rigs are conceived or arranged in
and one wishes to have 3-D. This is the case,
two ways when they can be turned around the
for example, for certain shots taken from a
helicopter.
optical axis: Integrated 3Ds camera
-The first, called zenithal, favors the protection
RIGS AND STEREOSCOPIC of the mirror. The vertical camera is placed
Also appearing on the market are professio-
nal video cameras that record two views of
Semi-transparent mirror rigs
CAMERAS The cameras are placed vertically at a 90° an-
under the optical axis and directed upwards. the same scene through two optical axes in-
This arrangement allows for a simpler protec- tegrated within the same camera body. This

A
gle to each other and find themselves in opti- tion from lights interfering with the mirror.
“rig” is an adjustable support on which type of wholly integrated technology has
cal concurrence in a semi-aluminized mirror.
movie cameras can be arranged and -The second, called “crane position”, has the the advantage of limiting problems of dis-
This mirror, also called a “beam splitter”, al-
set up. There are two types according lows one camera to film through the mirror
to whether the cameras are placed side-by- and the other to film by reflection on the re-
side or as a mirror (beam-splitter). It is the flecting side of it.

Binocle
need to obtain or not a small or large inter- For this process, the transmission through,
axial distance that determines the choice of a and the reflection from, the mirror is equal
rig. If a large inter-axial distance is desired, to 50% each so that the loss of a diaphragm,
usually for helicopter or distance shots, a si- induced for the camera filming through the
de-by-side rig is used. mirror, is identical, or almost so, to the other
The sophistication of side-by-side or mirror one. The mirror thus permits filming with an
rigs is not linked to the device but to the inter-axial distance equal to 0 and thus be
definition made by the manufacturer. Certain freed from the problems related to the space
manufacturers make rigs that are versatile used by the cameras and lenses.
and are simply assembled differently to be It is thus possible to reduce the inter-axial
installed either as a side-by-side or mirror distance to 0 mm, which allows for the ca-
rig. meras to be perfectly aligned. The mirror rig
mainly permits small inter-axial distances,
Side-by-side rigs indispensable for filming close to the sub-
Two identical cameras are placed next to ject, which is the case in a large majority of Simplified schema of a semi-transparent rig

22 23

METHODS FOR ALIGNING AND

DR
SYNCHRONIZING 3Ds CAMERAS

A stereoscopic shot requires paired equipment (cameras and lenses) and
the stereographer’s utmost attention to the adjustment of the rig during shooting.
Not being able to have perfect alignment once the shooting starts, it is necessary
to continue working thanks to real-time correction algorithms for television, and
in postproduction for cinema. It is a question of bringing to the viewer two images
perfectly aligned vertically pixel by pixel.
Factors provoking troublesome vertical disparities:
- Difference in height of the optical axes (whether they are parallel or cross verti- Two examples of integrated 3-D cameras
cally)
- Difference in the placement of one camera in relation to the other parity between the two stereoscopic points To this end, technical creation industries have
- Difference, even minimal, between the focal lengths of the cameras and the of view to be superimposed, especially pro- seen, since the end of the 90’s, the rise of
focusing of the lenses blems of alignment, differences in focusing, motion control rigs capable of being precisely
- Different optical aberrations between the lenses of zooming, etc. However, the integrated paired electronically via a computer. More
- Trapezoid effect resultant in vergence on the object. 3Ds camera has the drawback of generally recently, 3Ds image analysis instruments
It’s a question of minimizing as much as possible these disparities when offering only a single setting for the inter- have arrived with, notably, the possibility
filming. On a side-by-side rig, the optical axes are placed in parallel aiming at axial distance and a limited adjustment of of making a precise real-time drawing
infinity, and a test card is used at 3 meters to verify that the altitudes have been vergence. These restrictions result in heavy out of problems of disparity between two
respected. On a mirror rig, the same procedure is used in placing the inter-axial constraints when filming a close-up scene in stereoscopic images and correcting them in
distances at 0. In both cases, test cards and a 3Ds monitor is used to precisely 3-D, and, more generally, does not allow for real time.
adjust the alignment, if possible to the size of a pixel. complete control of stereoscopic production All these factors make it so that today one
Likewise, that the colorimetry and luminosity of the cameras are identical through adjustment of the two parameters sees hardware and software solutions appear
should be watched over closely. Certain tools on the 3Ds monitor allow this to be inter-axial distance and vergence. from several components manufacturers
precisely monitored. for processing 3Ds shots in real time,
Synchronization of the camera sensors at the pixel level is indispensable. The non- Progress of 3Ds rigs and monitoring of disparities which allows certain corrections—such as
respect of this rule on moving images leads to visual fatigue by the viewer. Whatever the talent of the stereographer, vertical disparities, detection of rotation and
For the timecode, the cameras must be connected with one camera as TC involuntary disparities between the two trapezoid deformation—to be automated.
“master” and the other as “slave”. viewpoints of the cameras remain. It is Certain software also permits automation
As to scanning, putting one camera in Genlock master and the other indispensable to correct them when filming of the inter-axial distance and convergence.
in Genlock slave is not precise enough, except for certain miniature cameras or live. For fiction, these defaults can also be Solutions of this kind, conceived in France or
cameras conceived for this purpose. It is preferable therefore to have recourse to an corrected on the film site so that rushes can elsewhere, which allow these problems to
external synchronization camera body in Tri-level Sync, connected to the camera be viewed comfortably. In postproduction, be corrected or parameters to be monitored
Genlock inputs. if these defaults are very minor and do in real time, are already being used on film
Cameras without the Genlock input, such as digital photo cameras, are not bother the editing process, certain sites, particularly in live sports productions.
problematic. Certain 3Ds monitors allow, however, scanning to be viewed and to stereographers prefer to usefully make these 3-D correctors are also being integrated
try “manual” or “on the fly” synchronization by triggering the cameras simulta- corrections during the “3-D calibration” into visualization tools used in fiction or
neously. But this does not provide the indispensable guarantees need for a feature stage. This consists in adjusting the 3-D for capturing in order to judge the effects
film project or for live television. values for a shot or between two shots. It obtained without mental strain for the team.
ensures the consistency of the 3-D effect for
the length of the film.

24 25

LIVE MULTI-CAMERA 3Ds A modification of workflow should be consi- As a general rule, it should be noted that all 3Ds POSTPRODUCTION:
dered in the 3Ds production unit. In fact, each the actors who intervene in the live stream
CAPTURING stereoscopic source should undergo 3-D cor- should have 3Ds visualization: slow motion CORRECTIONS AND WORKFLOW
rection before being integrated into the grid operators, synthesizer operators, special ef-

A
Live multi-camera 3Ds capturing implies ac- or the mixer. This correction can be multiple fects technicians, etc. ccomplishing postproduction in 3Ds
tions in real time in order to preserve 3-D ho- and specific to each rig. It encompasses the is not very different from traditional
mogeneity. The first action consists in analy- flip image in the case of a mirror rig, geo- The difficulty of live 3Ds multi-cameras today postproduction, if only that it is neces-
zing new actors and organizing them into the metric correction, compensation for opti- is not only related to technical problems, but sary to manage file volumes twice as large.
production team’s hierarchy. The director is cal centering, etc. Whatever the nature and also to the calling into question of the pro- It is also necessary to ensure the continuity
the one who conceives the stereoscopy of a number of corrections applied, they should duction. It is quite clear that the acceptance of depth in 3Ds scenes, by watching over the
project that puts the stereographer and all be done in a very short time since this period of 3-D is directly dependent on the quality of sequence of their stereoscopic ranges and
the technicians to work. comes before mixing. Today, this processing the stereoscopy, but it is also necessary to by correcting the disparities between the left
A single stereographer guarantees the qua- time varies from 40 to 120 ms. provide the necessary information for a good and right images. When filming fiction, these
lity of 3-D capturing. He works in concert The quality and number of 3Ds visualizations narrative. Due to this fact, live broadcasting defaults can also be corrected, of course,
with the director of photography. He is assis- is a key element of a 3Ds production tool. is a dangerous exercise in 3Ds. For example, before reaching the postproduction stage.
ted by the convergence operators who are in Two cases are possible for the production: to if one wishes to bring information captured However, if these defaults are very minor
charge of the stereoscopy of one or two axes. propose a multiviewer which includes all the by a 2-D camera into a 3Ds stream, one can and do not interfere with editing, certain
He continuously regulates the inter-axial sources as well as the program and preview resort to a real-time 2-D/3-D transfer, or if stereographers prefer to make these adjust-
distance and vergence according to the vain 3Ds. This requires a certain vigilance in or- one wishes to provide a score or a time, one ments during 3-D calibration, or, in any case,
riations of frame, focal length and focus. der to exclude from the mosaic any camera would proceed by inserting 3Ds elements these 3-D adjustments are reviewed in terms
By analogy with photography, it is necessary being adjusted. The second option consists into a 3Ds image. These techniques are to- of consistency over the length of the film.
to integrate this team into the live production of keeping the sources in 2-D visualization lerable for the viewer as long as they are
unit in order to make discussion easy and and to put only the program and preview in mastered.
3Ds.
Viewing is at the heart of the process of 3Ds editing
guarantee a consistent program. The postproduction of 3-D images requires

26 27

OPTIMIZING LIVE 3Ds COSTS IS POSSIBLE ! ting tools, this is rarely the case.

copyright : Technicolor
For this reason, during “offline” editing of
3Ds films or programs, it is imperative to
Concerning the additional costs related to 3Ds capturing, the figure of 25 to 30%more than view 3Ds edited shots and to validate them
2-D capturing is often put forward, but according to those in charge of AMP VISUAL TV these regularly on a stereoscopic screen having
figures are relative. The additional costs greatly depend on the length of the capturing itself, a size near to the one that will be used for
and AMP VISUAL TV hopes to have, in the end, enough production streams in 3Ds to be able the final showing. In this way it is possible to
identify at once the visual discomfort resul-
to have economies of scale. What is most important today is to find methods for reducing the
ting from the non-corrected stereoscopic
installation and preparation times. If it is possible, for example, to have two cameras/lenses disparities or the different 3-D values used
used only for 3Ds capturing, it would then be possible to considerably reduce the preparation in different shots. Example of correcting geometric
time before shooting. But for this to happen, it would be necessary to have a greater volume of If possible, it is even desirable that the conti- disparities
production than there is today. nuity of the shots be studied during the pre-
paration phase of filming in order to make
the 3-D space consistent before and after
a return to more rigorous methods of work after editing, during or just before the stage the transition by reducing to the maximum
than before, and to a more precise definition of calibrating the 3-D of the film. It is done by the differences in distance and convergence
of each person’s task, from preparation of a stereograph technician having experience of the cameras. Nevertheless, adjustments
filming to postproduction. These precau- with the correction tools, under the direction are often to be planned in postproduction
tions, taken before the various operations, of the stereographer who has already wor- through horizontal transfer of the left and Flare: a source of visual discomfort
limit the alternation of inconsistent 3-D se- ked on it during filming. right images.
quences. The change in location of the space to whe-
However, even if a stereographer has per- Continuity of shots during editing of 3Ds images re the viewer is attracted, from one shot to
fectly adjusted the rigs and taken care to stereoscopic range at the end of the prece-
The rhythm of shots edited in stereoscopy another, can also cause visual discomfort.
limit the “bad” connections between shots, ding plan consistent with the beginning of
is logically slower than that of editing of 2-D Thus the necessity of an appropriate script
the editor, will in all likelihood, as in the case the next one.
images, for when there is a change of shot, it and a specific “3-D calibration”, making the
of 2-D, change the editing order that was is necessary for the brain to have one second
planned at the beginning. Due to this fact, to appreciate the new universe to be explo-
it will be necessary to again verify the conti- red. Learning to watch 3-D by the viewer Adapting convergence and editing:
nuity between shots during each viewing of can, however, reduce the accommodation 3-D continuity between two shots
the 3Ds already-edited rushes. time necessary for each change of shot. But
Furthermore, following the planned post- as of today this is only a supposition. In 3Ds, convergence between two successive shots must be adjusted. Usually, the convergence shots
production method and the production bud- In order to limit mental discomfort and re-
get, it might be useful to take some time to will be brought closer during cropping in postproduction if they are too different. Otherwise, the
duce adaptation time between each shot,
correct the involuntary disparities between it is also possible to use the “3-D dissolve” viewer may feel visual discomfort. The time this adjustment takes can also vary. Experiences that
images. This would mean to correct the- method which smoothes the passage from took place during the production of Legend of the Guardians: The Owls of Ga’Hoole (2010) brought
se geometric and colorimetric disparities one 3-D shot to another when the 3-D ef- to light the fact that the softest transition in 3Ds is obtained when the vergence of entrances into a
between the left and right images such that fects are obviously too different.
each of the homologous pixels is situated at
shot is progressively adjusted according to the previous shot. In this film, the length of the vergence
Because of these constraints, it would be
the same altitude. It must be kept in mind more practical in the absolute for stereos-
adaptation time between shots varied from one sequence to another from 8 to 40 images. Further-
that these corrections, necessarily made for copic editing to take place “online”, directly more, rather than making the depth characteristics of each shot exactly correspond to the level of
a live capture, should become a postproduc- from already-corrected stereoscopic ima- the cropping point, it was noticed that for most of the sequences, it was possible to eliminate only
tion operation for fiction. ges. However, for reasons of visual fatigue 50% of the difference in depth between shots, and to leave a slight upsurge in 3Ds when cropping.
This operation could be made on the rushes over time, but also for the sake of economy
before editing, but today it is generally done Of course, nothing prevents one from using a brutal 3Ds connection between shots to provoke a
and/or the non-availability of online 3Ds edi-
particular sensation in the viewer.

28 29

Hunting down contrasts that are too strong the parameters of each element in real time,
Two types of 3-D according to objects entering into Speed, acceleration and the trajectory of the Since 3Ds broadcasting systems (polarizing and, if possible, on the size of a screen clo-
filters and glasses, liquid crystal glasses, sest to that of the definitive broadcasting size
the shot cameras
etc.), do not have an absolute separating in order to take into account the magnifica-
The viewer feels visual discomfort if one The movements should preferably be fluid
power, each eye more or less perceives a tion factor.
of his eyes sees part of an element that and slow, with slow accelerations and dece-
phantom of the image destined for the other The special effects person must be trained in
the other eye does not yet see coming into lerations and trajectories in wide curves. If
eye. It is thus necessary to limit contrasts 3-D and know how to measure and precisely
the shot from the side. A wise compromise the scenario imposes rapid effects far from
that are too strong between foreground place the different elements of a composite
between the intensity of the 3-D effect and such recommendations, reduced 3-D should
and background elements according to the image in a stereoscopic 3-D scene. The pla-
the speed of entering into the shot will need be chosen: in the extreme, the optical axes
intensity of the 3-D effects and the created cing of these elements must necessarily be
to be determined. This does not pose a pro- of the left and right cameras can be blended
horizontal disparities. very precise. The objective of the visual effect
blem, on the contrary, when this entering little by little so that one films in quasi-mo- is to modify the perception of the 3-D space of
into the shot is made either on the plane of noscopy, thus preserving visual comfort. It each element to be integrated by using:
convergence (plane of the screen) or verti- must be added that rapid movements, espe- Adjusting the light for popping-out effects • The horizontal transfer of the left and right
cally (from the top or bottom of the screen), cially towards the observer, can give rise to To reinforce the sensation of immersion, the images, modifying the convergence point and
for in these cases both eyes see the same ocular fatigue. visual field of the viewer must be filled. The thus the depth location of each object in the
element appear at the same instant. best thing is to progressively dim the zones scene (“horizontal shift”).
of the popping-out object that cut the edges • The resizing of the images after the removal
of the frame. To do this, a method consists of the parts that are not common to the left
The case of a film having “heavy” digital visual effects? in reducing the light of a scene progressively and right views (“crop” and “resize”).
towards the edges of the screen. The volume • The correction of the vertical disparities,
Concerning geometric and colorimetric defaults, there is a tolerance threshold beyond which the of the filmed scene then seems to be entirely especially trapezoid effects (“keystone”) due
viewer gets tired, and it must be admitted that there is a certain imprecision in the correction of centered on the viewer. to the convergence of the two cameras.
these defaults. On the other hand, the fine tuning of different elements of a composite image must
be accomplished with “zero tolerance”. It is for this reason that the SFX supervisor takes charge of Matte painting in 3Ds
all the image’s 3-D manipulations: Matte painting is a technique that can always Adjusting convergence in postproduction
1 - Correction of geometric and colorimetric defaults as well as putting the image filmed in be used in 3Ds for distant scenery, in bac- It is important to note that 3Ds convergence
can be adjusted in postproduction and espe-
convergence in simultaneous parallel mode. kgrounds toward infinity, where the 3-D set-
cially for special effects sequences contai-
2 - Creation and assembly of the composite image’s elements. tings are closest to human vision and without
ning visual effects (see box). In this case,
3 - 3-D fine tuning of these elements, in particular in the case of assembling live and computer- real perception of 3-D. However, it is indis-
filming often takes place in “parallel” mode,
generated images. pensable in this case to create a matte pain-
and convergence is done in postproduction,
4 - Putting the composite image in convergence, and 3-D calibrating for the projection screen. ting that is very precisely integrated into the
which is much simpler.
Important note: when there is filming for a composite image (special effects), it is not recommended “composite” 3-dimensional space.
It is no longer convergence that is adjusted,
to correct the defaults of the 3-D image before the special effects, because this step produces a The distinguishing of composite stereosco-
but the position of the image according to
destructive “filtering” that harms the quality of the special effects at the end. pic 3-D elements by the viewer is in fact is
the depth: the whole image being brought
Besides the SFX supervisor, a certain number of positions can be identified whose work methods much more keen than within a 2-D image.
closer or set farther back, and this without
should be adapted to 3-D images: An imprecise composition of a 3-D image will
deformations. It should be noted that a
-rotoscopy be immediately seen as incoherent by the
viewer, and even more so when the size of the small part of the side of the images will be
-tracking lost, while the vergence on the set will lead
-compositing broadcasting screen is large.
to a rather slight loss at the top and bottom
-matte painting and all work of changing the palette of the image during the elimination of the
-computer-generated graphics Compositing in 3Ds trapezoid effects.
The computer graphics artist in charge of this work must be labeled “3-D ready” in order to In order to precisely composite the different
guarantee the quality and productivity of his work. elements of the final scene, it is necessary to
Management of 3Ds files
acquire the tools for viewing and modifying
More and more shootings in 3Ds use cameras

30 31

that record in “file mode”. It then becomes 2-D/3Ds CONVERSION illusion that he sees from two different view- steps and intervention of a specialist remain
very important to recover the X and Y infor- points. indispensable without which the possibility
mation from the cameras to integrate them THE NEED FOR KNOW-HOW exists of providing insufficient visual comfort.
into the metadata. In 3Ds postproduction, one

C
Filming in 2-D while respecting 3-D constraints A number of disparities between the image’s
of the main questions is the management of hoosing conversion according to filming two viewpoints must be corrected manually.
However, when filming in 2-D in view of
storing these files, which becomes crucial. Likewise, when distant objects are blurred in
conditions converting to 3Ds in postproduction, it is im-
The best thing is to opt for regular backups the original 2-D image, the infinite in the re-
Today, several American blockbus- perative to film while already extrapolating
of the images as production proceeds, for the constituted 3Ds image must be redefined. It
ters shown in movie theaters in 3Ds are the the 3-D effects and in keeping in mind the
quantity of data to be processed is doubled. is also necessary to recreate some 3Ds ma-
result of conversion of 2-D images into 3Ds constraints pertaining to 3-D. This is how the
The method that consists of editing online terial in zones where the 3-D effect is pro-
in postproduction. The main reason is econo- beginning shots, when characters are often
directly in 3Ds is most often used for short nounced.
mic and comes from the fact that producers, cut off during filming in 2-D by the edges of
programs (publicity, corporate films, docu-
in front of the obstacles related to filming in the frame, can prevent using popping-out ef-
mentaries, etc.). This method assumes that
3Ds, prefer to respond to the demand of the fects in the scene. One can slightly cheat to Automatic 2-D/3Ds conversion: an aberration…
there are enough editing stations sufficiently
3-D digital movie theater market through have a slight popping-out effect, but with the Most of the current automatic conversion
powerful to process 3Ds images in full reso-
2-D/3Ds conversion, often easier to manage risk of generating the beginning of visual dis- systems provide stereoscopy processed only
lution and in real time.
when it has to do with films integrating a comfort in a section of the public, especially in the sense of depth and immersion without
large number of digital special effect shots. if this choice is repeated throughout the film. popping-out effects. These machines work
Longer editing For example, for the needs of creating 3-D it More generally, shots could have superb with the help of algorithms resting, among
The editing of a 3Ds project is accomplished is possible to cut out persons who have been popping-out effects provided that the frame other things, on a principle of color recogni-
in certain cases from a single point of view, filmed with a green background and to re- is kept “airy”, which requires the cameras to tion, with distribution organized on the fact
for example, the left image. To appreciate process scenery in 3Ds that has been trac- pull back or to have the foreground charac- that warm tones are mostly in the foreground
the consistency of the final editing, it will be ked by a virtual camera. Usually, this sce- ters moved back or recentered. of a scene and cool tones in the back far
necessary to make regular conformations in nery will have been stretched out in matte Certain transitions and dissolves already in away. The only problem with this so-called
3Ds, allowing the editor to modify the struc- painting with computer-generated images place in a film or TV program shot in 2-D are “natural” rule is that it includes many excep-
ture and rhythm of the film according to the of creatures inserted into the scene and, in going to have to be adapted to the constraints tions, for the simple reason that one is in an
3-D effects obtained. this case, the stereoscopic processing of the of the transitions between 3-D spaces, or artificially lit universe where, when someone
film can become more expensive and more else there will be volumes overlapping each has some clothing with a cold tone in the fo-
Lengthening the calibration time difficult to manage than a 2-D/3Ds conver- other without any spatial coherence. reground, warm tones may be observed in
The calibration of the 3Ds calibration screen sion in postproduction. Consequently, putting 2-D images into 3-D the background. Due to this fact, automatic
should be made according to the technology of It is also possible to mix techniques within is a technical process demanding the grea- conversion generates many visual artifacts
the glasses used (passive or active). Besides, it the same film. Such “composite” projects test care to prevent visual discomfort or and ambiguities. These ambiguities are
is difficult to plan the calibration (colorimetry then consist of shots filmed in 3Ds and even rejection by the viewer, all the while going to quickly turn the viewer away from
and density) with 3Ds glasses being used for others converted from 2-D to 3-D when the respecting a certain realism in 3-D. During this type of image which gives a truncated
numerous consecutive hours without having sequences are easier to process by 2-D/3Ds 2-D/3Ds conversion, the 3-D should be set view of reality.
ocular fatigue. Usually, calibration is done in conversion rather than by filming. This is the up essentially to ensure the visual comfort
2-D. Then, during viewing in 3Ds, a correction case, for example, of sequences filmed with of the viewers, with beautiful window effects …needed for 3Ds program processing
is applied on the viewing screen (LUT) to com- long focal lengths in stereoscopic mode that but with few effective and comfortable pop- However, TV broadcasters will be quickly
pensate for the effect of using glasses when can give the effect of successive “slabs”. ping-out effects given the difficulties in set- constrained to use automatic conversion
viewing the images. This flattening effect, due to filming with ting them up. from 2-D to 3Ds in order to accomplish pro-
At this stage, it is necessary to adjust the ste- long focal lengths, is especially irritating gram continuity between different programs
reoscopic range of the film, also called “3-D when filming a close-up portrait shot. Then Correcting disparities in 3Ds sufficiently smooth and acceptable to
calibration”, or to modify the colorimetric the simplest and most economical thing to 2-D/3Ds conversion, accomplished automa- the viewer, such as, for example, between
calibration according to the 3Ds perception do is to film the sequence in 2-D and then tically with the help of postproduction ma- trailers and publicity that have been made
constraints. add contours to the face using rotoscopy of chines, is certainly becoming more and more with different 3-D settings.
the person such that the viewer is given the efficient, but it is no less clear that manual

32 33

Animation in computer-generated graphics, matte truly have two viewpoints of the objects (one cherche (France), has the goal, among other - flicker, when the left and right views are al-
for the left eye and the other for the right things, of corroborating the results of labo- ternately shown too slowly
painting and stereoscopy
eye). For this type of precise shot, a classic ratory tests with a more restricted panel of - when the colors or brightness of the left
All the shots of a film or a TV program using
output of a film or TV program in 2-D can- 3Ds viewers. A vast study using the general and right views of the same element in the
matte painting for background scenery
not be recovered to be used as a basis for public should be published during 2012. A image are too unequal
should be studied before postproduction if
3-D. It is thus necessary to take up again all small sample of people has already been put - the contradictions between monocular and
conversion from 2-D to 3Ds is desired. It is
the 3-D files used in compositing with matte under visual constraints that are supposed to binocular indications of depth, particularly
especially indispensable to determine which
paintings and digital visual effects. provoke problems in order to objectively de- those that result from a default in synchro-
matte-painting elements should be recrea-
fine certain indications of discomfort. Some nization during rapid side movements, or
ted by 3-D computer-generated graphic
examples are: loss of visual acuity, increased “window violations” that make it so that an
models in order to obtain a second camera
movement of the eyelids, contraction of the element of the image is both, because of its
viewpoint that is visually consistent. VISUAL COMFORT WITH 3Ds pupil, slowing of the change in orientation parallax, seen in front of the window, but hid-
For matte painting of background scenery to-

I
of the ocular axes, loss of binocular fusion, den by it
wards infinity, it is possible in fact to “cheat” t often happens that viewers of 3-D films etc. - images that are too complicated (high spa-
by creating 3-D through compositing. Our complain of visual fatigue, nausea or dis- Among the diverse causes of visual discom- tial frequencies)
human perception of 3-D towards infinity comfort. This dissatisfaction can have two fort with 3Ds, most of the studies concentra- - movements that are too quick by certain
is, in reality, almost nonexistent (due to the main causes: ted on the visual disparity between the two elements in the image so that the viewer
small distance between our two pupils com- -defaults during the making and/or broad- viewpoints in relation to convergence and does not have time for binocular fusion to
pared to the distance separating them from casting of 3Ds content accommodation. While in natural vision our take place before the image completely
observed objects). It is thus possible to easily -rejection of the artistic means (rejection of ocular system converges and accommodates changes
fool the brain through this type of composi- its use or disinterest) on the same object, the viewer of a 3Ds film Each of the visual discomfort factors acts in
ting operation. In-depth studies on the main causes of visual accommodates on the plane of the screen, a very different way from one person to ano-
For matte painting and digital visual effects discomfort with 3Ds for the viewer are made but converges towards the objects that at- ther. When creating 3Ds content, one must
closer to stereoscopic virtual cameras, one from samples of representative groups. tracts his attention and can be situated just be very vigilant concerning each of these pa-
must, on the contrary, model the scenery in The 3-D Comfort & Acceptance project, fi- as easily on the plane of the screen, in front rameters.
3D computer-generated graphics in order to nanced by the Agence Nationale de la Re- of it or behind it. We hope that the current scientific studies
will establish clear physiological thresholds,
Wide variability according to the person and in so doing define more precisely the
RELATION BETWEEN ACCOMODATION But the effective limits noticed are extremely possibilities of creating 3-D that is agreeable
AND VERGENCE IN 3Ds variable from one person to the next: a small to watch.
proportion of people are less sensible to dis-
parities in the lower limit of this “comfort
The zone in blue corresponds to the zone of ordinary,
zone”, whereas another small proportion
sharp binocular vision of a normal subject (ZVBSN). This will only feel discomfort more than five times
diagram shows how in natural vision the demand for ac- the higher limit.
The most well-known causes of visual dis-
commodation and vergence in the human ocular system
comfort with 3Ds are mainly:
are covariant in a way that can be predicted according to - divergence of the ocular axes (heteropho-
a linear relation (demand line). In 3-D vision (3Ds), this ry)
- amplitude of variations of angular parallax
coupling is no longer in perfect concordance and can lead
- vertical disparities
to convergence accommodation conflicts causing stinging - “orphan” elements in the image, that is, vi-
and/or irritation of the eyes, headaches, etc. (asthenopic sible in one eye and not the other, especially
if they attract the attention (very bright, for
symptoms).
example)

34 35

ter sends a synchronization signal,
adjusting the synchronization. reflected by the screen’s canvas,
To not tire a person’s vision, it is which easily allows all the seats to be
important that the synchroniza- covered. Several transmitters can be
tion of the display of the images, placed in large theaters.
using the switching of separator The classic movie theater screen can
systems for each eye, is perfect. be kept. The less the gain is high, the
These adjustments are to be made better the result will be in terms of
for each installation. quality (in particular the uniformity of
Polarization of light the luminosity).
Polarization of light Active shuttering The viewer has “active” glasses. The-

3Ds PROJECTION FOR FILMS se glasses are equipped with liquid crystals
For the record, this principle has also been
theater. Uniform values of lighting are thus whose orientation can be modified. Accor-

T
used for a long time with 35mm projection.
he 3Ds projection technology currently A light polarization system is placed in front obtained that are very inferior to the recent ding to the eye and the synchronization si-
used in movie theaters is based on the of the projection lens. It is a question of al- standards for digital projection (ISO, Afnor). gnal sent by the infrared transmitter, these
sending of two distinct and successive ternately polarizing light in opposite circu- While acceptable for 3Ds projection whose liquid crystals will alternately block the
image streams, and a system allowing each lar states. All the images intended for the qualitative characteristics are not standar- viewer’s vision or let the light pass through.
eye to only receive information intended for left eye are thus projected with a polariza- dized, this particularity is very damaging Each eye thus sees the image intended for
it. tion in one direction, and those intended for for the quality of 2-D projections. It is thus it, and only this image (no “ghost” effect).
Three technological solutions are proposed the right eye with polarization in the oppo- recommended that theaters so equipped be The glasses should be supplied with elec-
to accomplish this separation: site direction. To accomplish this, systems dedicated to 3Ds. Furthermore, this method tricity to activate the LCD command circuits.
• Polarization of light of liquid-crystal filters (RealD solution) or is not very appropriate for wide theaters with These glasses are relatively expensive and
• Spectral filtering of light rotating disks (MasterImage) are placed in little depth. also quite fragile. They are usually “rented”
• Active shuttering front of the lens. A connection to the GP/IO To perceive the images separately, the viewer to the viewer for the time of the film, which
All of them come up against a real problem port of the projector ensures the synchro- wears glasses having a filter for each eye, requires logistics for recovering and clea-
of light output, for on average between 75 nization of these filters with the display of each only allowing light to pass according ning them. The main suppliers are Xpand,
and 85% of the light stream is lost in the the images. to its polarity. This filtering is not perfect, Volfoni and E3S.
systems (filters, glasses). The power of Problems could appear with the projection and each eye perceives a little of the image
xenon burners being limited (currently a window, since most glass integrates light intended for the other eye, thus generating Color filtering
maximum of 7000W), the dimensions of a depolarization layers for security (espe- a slight double image called “ghosting”. In The principle of this technology is to filter
lightable image using 3Ds projection with cially fire). order to get around this problem, DCP files by bands using a wavelength comb filter.
a single projector are necessarily limited The light is reflected by the surface of a type called “ghost images” can be used, that is, Two techniques have been developed, one
(14 to 15 meters in cinemascope format). of “metalized” screen. In fact, it is a screen integrating an inverse image of the parasite by Dolby (based on an Infitec patent) and
Beyond that, it is recommended to use two painted with a paint containing aluminum image for each eye. The glasses are sold to one by Panavision.
superimposed projectors. The luminosity of elements. This paint has the distinctive fea- the viewer (very low cost) who can keep and The Dolby solution inserts a color filtering
the images is a fundamental factor in the ture of keeping the polarization of the light reuse them for other projections using the wheel into the light path, between the lamp
perception of stereoscopic effects. when it is reflected, which is not the case same technology. and the matrices (three bands per eye, sli-
Another difficulty is the fluidity of the pro- with classic mat white canvas. The two main providers of this type of tech- ghtly offset for each). The rotation of this
jection. In order to improve this problem, On the other hand, these metalized canva- nology are RealD and MasterImage. wheel is synchronized with the alternate
and thus to improve the perception of 3-D, ses are very directional, that is, they reflect display of images for each eye. A colorime-
each image is projected three times (triple most of the light according to the effective/ Active shuttering tric processing must be made at the server
flash) in the space of one image, that is six reflection axes. Concerning the projector, there is no mo- level in order to reinforce the chromatic se-
images altogether in 1/24th of a second This results in a “hot point” whose loca- dification. An infrared transmitter is simply paration of the two images. This requires a
(three images per eye). tion varies according to where one is in the connected to the GP/IO port, generally pla- server able to do this processing.
The last difficulty in 3-D projection is in ced behind the cabin window. This transmit- The Panavision solution requires that the

36 37

right images is successive. Concerning to High Definition TV. It is what all the televi- ColorCode, who, thanks to a system of ma-
the projector, the same lens can be used. sion manufactures say and which explains naging the separation of colors, allows this
It is also possible to display two images the very rapid marketing of 3-D televisions already old principle of creating stereoscopic
“top-and-bottom” or “side-by-side” on in 2010, even if the content does not exist to- images on television to remain alive a bit lon-
the matrix (for example, if it is 4K). In this day in sufficient quantity. ger.
case, it is necessary to replace the lens by The other explanation for the rapid deve-
a double-lens system. This is the solution lopment of 3Ds televisions comes from the
taken by 4K Sony projectors. The polarizing
Polarized panels and passive glasses
improvement of the quality of the display The principle of passive TV3Ds glasses
passive systems are used in front of these technologies and in particular the frequency with a polarized screen surface placed on
double lenses. Recently, it is also possible of refreshing flat screens that allow this to a classic LCD panel is a technology used
to use active glasses. take place above 200 Hz. Given this fact, it is by certain consumer electronic brands. The
possible to propose quality 3-D while crea- principle of its display is based on a filter ha-
The technological battle between passive and acting a minimum number of undesirable ef- ving a circular polarization in one direction
tive systems fects for the viewer. for even lines and a polarization in the oppo-
3Ds cinema projection systems use a To really understand, however, the problems site direction for odd lines. In this way, each
Color filtering wide variety of different technological of displaying 3-D images on a television, it is eye only sees the lines corresponding to its
principles. It is the sign of an emerging first necessary to recall the main technolo- direction of polarization.
market which is in the process of clo- gies and to provide information on specific The main advantage of passive glasses re-
whole lens be replaced and a fixed filter be sing in today on a battle between the rules related to this new mode of television sides in the fact that no electronics are
inserted behind the back interior lens. This principle of passive projection of the type display. embedded in them since the glass only has
filter separates the light spectrum for each RealD or Dolby and the active system of the “static” polarization. These glasses are li-
eye (five bands per eye). type XpanD. This competition in the movie Exit anaglyphs ght, the colors are quite correct and there is
The classic movie theater screen can be theaters between two competing techno- As an introduction, we note that it is possible no flickering effect. On the other hand, with
kept. The less the gain is high, the better logies, each having its advantages and in- to visualize 3Ds images on a traditional televi- passive glasses each eye only receives a half
the result will be in terms of quality (in par- conveniences, will no doubt be influenced sion through the transmission of an anaglyph an image on the vertical resolution plane.
ticular the uniformity of the luminosity). by the arrival of new light sources in the image filtered in red and cyan. This already Furthermore, with this display technology,
The viewers are furnished with glasses that D-Cinema (Laser) projectors, which clearly old possibility is still used sometimes on when the head is inclined in relation to the
filter the light spectrum and allow each eye improve the quality of comfort when 3-D today’s televisions, but broadcasting in ana- television screen, diaphonic effects appear
to only view the images formed from the projection is used. Likewise, the techno- glyph has only a single advantage from now like in the technology used with “active”
color triplets intended for it. This filtering logical rivalry in movie theaters is likely to on: from the encoding of the video up to its glasses. Here, the circular polarization al-
must be very precise and requires the stac- be decided in the course of time by choices display on the television it is not necessary to lows the head to be inclined up to a certain
king of around 50 layers of colored plastic taken by large electronic manufacturers in have a specific broadcasting infrastructure. point, after which the polarization of the li-
film. Furthermore, as these chromatic fil- the mass television market, which little by On the other hand, this technique suffers ght causes loss of luminosity and colorime-
ters require the light to pass through at an little propose for both technologies more from serious limits concerning visual com- tric changes.
angle of 90°, the glasses are specially sha- and more refined and comfortable techni- fort. It intrinsically generates a loss of color
ped to allow the eye to move without losing cal solutions for the viewer. fidelity. In particular, depth information is TV3Ds with active glasses
the 3-D effect.
lost in those parts of the image having the In 2010, most of the television manufacturers
The glasses are very costly (less than ac-
same colors as the red-cyan filters. Even lost accepted a TV3Ds display technology using
tive ones). They are usually “rented” to the
sometimes is the capacity of the eyes to fuse.
viewer for the time of the film, which re- 3-D TV DISPLAY Also noted is discomfort due to the disparity
active glasses. This technology is based on
quires logistics for recovering and cleaning flat screens having at least a double refresh
of luminosity between the left and right filters frequency (even frequencies of 400 and 600
them. With the passage to stereoscopic 3-D, the of the glasses. Hz), that are combined with 3-D active glas-
development in terms of visual sensation for To be quite complete, it must be emphasized ses. The shutter frequency of active glasses
Displaying the images the viewer is much more pronounced than that the anaglyph process has been refined is synchronized with the display of images
For all these systems, the display of left/ during the passage of standard definition TV these last years by companies like Trioviz and on a flat screen. In this way, each viewer’s

38 39

eye only sees alternately the image that is principle is essentially the same as lenti- 3-D images to be made and thus limits de top-and-bottom in 720p50, 1080p25, 720p60
intended for it. cular network autostereoscopy in which a facto, on a technical and economical level, and 1080p24.
This technology is for the moment almost filter (the barrier) alternately distributes the production of content in this display
the only one to offer a 3-D image with full the points of view intended for each eye. In mode, at least for now. 2-D/3Ds conversion in the television set
resolution for television, even if companies contrast to lenticular networks, the lateral Dans la mesure où le relief créé par un réalIt
like RealD, JVC and Samsung now market positions for seeing the whole image are A single connection for 3Ds: HDMI 1.4a will be several years before TV3Ds natively-
a Full HD passive TV3Ds technology near all at the same distance from the plane of The new HDMI 1.4a (High Definition Multi- produced 3-D content becomes widespread,
to the Z-screen system used in movie thea- the image. It thus permits switching more media Interface) connection defines the new especially due to the high additional costs of
ters. This new approach should permit the easily from 2-D mode to 3Ds mode, which formats and timings for stereoscopic image production and postproduction generated by
polarization direction to be alternated on is not so evident with a physical lenticular information furnished to televisions through this new technology. From now on, television
the whole image, thus offering a Full HD network. readers and decoders. The HDMI 1.4a stan- manufacturers, far in advance of the native
polarized system for TV. These two technologies represent a strong dard, launched on 4 March 2010, has impro- 3Ds content production market, have deci-
The only drawback to the “active” glas- p o t e n t i a l for development ved several functions of the HDMI connec- ded to integrate real-time converters of 2-D
ses technology is, for in the future of tors, including the possibility of conveying images to stereoscopic 3-D images in their
example, when the TV3Ds display most of the stereoscopic 3-D signals used latest generation of televisions.
synchronization me- technology in so for broadcasting today. Among these si- This conversion function for the viewer is
chanism of the glas- far as they allow gnals, there are the “Compatible Frame” usually accessible with the help of a simple
ses is defective. In this the viewer to do modes based on classic image resolutions, remote control. It acts only in the sense of
case, double images without glasses. where the two images share the horizontal the depth of the image through a heighte-
(ghosting) can appear, Autostereoscopy, (side-by-side) or vertical (top-and-bottom) ned sensation of immersion in the scene.
which is a nuisance as 3-D immedia- resolution. 2-D/3Ds conversion in the television set is
for the viewer. The tely visible without These broadcasting modes are intended, not conceived so as to produce objects that
problem also results glasses, already above all, for the TV transmission of 3-D pop out from the screen. Such automatic
in diaphonic image represents, mo- content, while the existing infrastructure 2-D/3Ds conversion in the home also causes
information of the left reover, a convin- can be used for the transmission in high-de- frequent artifacts, exaggerations of 3-D on
eye onto the right eye cing medium for finition (HD). Thus, where HDMI 1.3 allows parts of the image that may quickly gene-
and vice-versa, while the Panasonic P55VT50 3Ds television marketing and only one HD image to be transmitted, which rate visual fatigue in the viewer.
periods of each eye being communication ap- means all diffusion modes of type side-by- So, even if it has yet to be proven by an
open for too short a time plications: public side, top-and-bottom and checkerboard, epidemiological study, it is already known
can lead to a reduction in places, dynamic dis- that is 3Ds with loss of resolution, the HDMI with certainty that such on-the-fly 2-D/3Ds
luminosity. play, museums, airport halls, etc. 1.4 standard allows the transmission of two conversion of video sources that have not
On the other hand, autostereoscopy is for HD images, or 3Ds in full HD. been conceived using native support or in
Autostereoscopy now still reserved for events and corporate As examples, there are several image for- postproduction to be broadcast in 3-D (ra-
The principle of autostereoscopy is based communications or even as a display inten- mats compatible with the HDMI 1.4a interfa- pidity of editing, continuity of shots, para-
on two distinct processes: ded for a single individual (telephone and ce: 3-D field alternative (interlaced), frame sitic objects in the foreground, etc.) leads
- An LCD panel in front of which a video game consoles, smartphones, etc.). packaging (top-and-bottom format with full to enough parasitic effects so that the in-
lenticular network is placed so that an ob- It has, in fact, two major constraints: the resolution by view), complete alternation of terpretation of the 3-D by the human brain
server only sees, for each eye, one column viewer must be in front of the screen and/ lines, half side-by-side, full side-by-side and becomes quickly uncomfortable with heada-
of pixels at once. One simply has to use a or to not move too much in front of it to keep the principle of 2-D broadcasting plus depth ches eventually occurring.
matrix of two images together (one pixel out a comfortable stereoscopic effect. Further- metadata. It should be noted that HDMI 1.4
of two) so that each eye receives a different more, to allow the viewers to be freely pla- works with 3-D flat screens that support Adjusting the intensity of 3-D in the television
image in real time and so the stereoscopic ced in front of an autostereoscopic screen, it definitions and broadcasting frequencies as In so far as 3-D created by a director is more
effect is produced. is necessary to display eight different came- different as 720p50, 1080p25, 720p60 and or less comfortable for the viewer accor-
- The parallax barrier is the other ra views enmeshed in a single image. This 1080p24, but also horizontal side-by-side ding to his own visual perception and the
important autostereoscopic technology. Its therefore implies the production of complex mode in 1080i50 or in 1080i60, as well as

40 41

size of his screen, certain television and ADAPTING 3-D TO THE SIZE OF to the camera. It is thus possible to recali- to broadcast the film, because the content
3Ds decoder manufacturers are beginning brate all the stereoscopy of a production by created for a large screen will always be
to offer manual or automatic adjustment THE BROADCAST SCREEN making a simple side offset of the stereos- viewable on a smaller screen, even if the 3-D

T
of the depth of 3Ds scenes. Whatever the here exists several phenomena and copic images. will be more subdued, whereas in the oppo-
electronic system used, it is clear that such rules that govern the viewing of ste- site situation, content planned for a small
adjustments made by the viewer cannot be reoscopic 3-D images that we have Calibration of 3-D according to the size of the screen cannot be viewed on a large screen
proposed at a level of 3-D going beyond that summarized in the following diagrams, without provoking visual discomfort due to
projection screen
conceived at the beginning of the program’s and which should serve as a guiding thread divergence.
If one considers that the perception of 3-D
production, the minimum level being, of when creating the preproduction of a TV Therefore, a film planned for a screen of 20
depends on the size of the image and posi-
course, a flat image in 2-D. program or film. meters at its base can always be projected
tion of the viewer, the adjustments of 3-D
depth and popping out must be fine-tuned on a screen of 2 meters at its base, at the
The placing of subtitles and their presentation Depth represented on the screen: a “scenic box” according to whether one is watching a mo- cost of a weaker 3-D effect, but the opposite
The real depth and the depth represented creates too large left-right horizontal spa-
on the air vie theater or television screen.
on the screen in a 3Ds project are very dif- In a stereoscopic image, space is subdivi- cing so that the 3-D can no longer be fused,
The placing of subtitles and graphic
ferent things, and trying to show the actual ded into two volumes, one part in front of necessarily creating visual discomfort.
elements in a 3-D television program So if the same film is projected on a larger
dimensions of the subject on the screen is the screen and the other in back. Between
is a technological and artistic feature a completely vain undertaking. The depth screen, the distance between the “homolo-
large movie theaters and the living room of
that must be taken into account before shown on the screen is called the “scenic the average man for example, this propor- gous” points of the “distant objects” (ele-
the program is to be broadcast. It is re- box”. It obeys laws belonging to represen- tion between what is in front and what is in ments discernible the farthest towards infi-
commended that scenarios should be tations in movie theaters or on a television back changes. nity) is going to increase quasi-proportional
built for the placing of subtitles and/ and above all to the image shown (see dia- For all that, is it necessary to “calibrate” the to the base of the screen and thus go beyond
or graphic elements based on an ana- grams below). 3-D according to the size of the television or the average inter-ocular distance of human
lysis of human perception so that they Nevertheless, one must be vigilant, becau- cinema screen? Is it necessary to adopt a beings (65 mm).
se sharing foreground objects (popping-out median screen value for both these broad- The limit resides in the fact that the homo-
do not interfere with the 3Ds effects in
effects) and backgrounds (depths, distant cast vectors? Is it also necessary to take into logous points of objects located the farthest
the program’s scenes. Two options are
objects) by specifying a precise position for account an “ideal” distance of the viewer in towards infinity on two stereoscopic images
possible: either the subtitles are always the plane of convergence is only valid for a relation to the television screen? should not be more than 65 mm apart, on
inserted in front of the video content given screen size. In particular, this plane pain of provoking ocular divergence in the
One of the first basic rules consists in ca-
(fixed depth positioning) and the rest of of convergence makes no sense with giant librating the 3-D of a TV program or a film viewer.
the scene is sent to the background, or screens of the IMAX© type, for with them, according to the largest screen size able
their position dynamically adapts to the all the represented volume can only be in
stereography of the program so that the theater itself. In fact, 6.5 cm on the basis
they are always in front of the content, of a screen of 30 m is so nonsensical that it Principle of the Floating Window
is no longer taken into account (0.2%, that
while avoiding too great of a parallax
is, 8 pixels out of an image of 4K). The im- The floating window is an arrangement of the stereoscopic image such that the window
between the scene and the text.
portance and exact positioning of the plane is seen in front of the screen. It is obtained by the creation of a virtual screen frame that
of convergence decrease steadily with the is closer to the viewer than the real screen frame in order to entirely or partially hide
increase in the size of the screen. the violations of the stereoscopic window.
Inversely, small screens are perfectly si- The floating window appears as black vertical bands on the left (in the left view) and
tuated in terms of depth, and consequently
on the right (in the right view), which reduces the horizontal visual field all the more a
adapt to most of the image in back of the
virtual frame close to the viewer is desired. The floating window can be dynamic and
surface. This leads to a stereoscopic win-
thus constantly change to adapt to the scene being broadcast. It is frequently used in
dow that is more in front within the volume
animation films.
of the scenic box, just as during filming
convergence would be made a little closer

42 43

Relation between 3Ds parallax and the size of the screen
The CST, in collaboration with Alain Derobe, created a table establishing the average settings of
distant objects viewed on different screen sizes: by percentage, fraction and screen discrepancies,
measured in cm or in number of pixels of the 2K image.
Experiments have been made on the rendering of distant objects for them to be as agreeable as
possible for an audience of spectators at common viewing distances.
For example, if a film is calibrated for a screen of 10 m at its base, thus involving distant objects
offset by a maximum 16 pixels, it will be necessary, aided by a general offset of the image, to reduce
them by 10 pixels for a presentation valid for a screen of 26 m at its base; and inversely, to increase
the offset by 11 pixels to go to a screen of 4 m at its base.

Explication du tableau
Except for small screens, the table is based on the viewpoint of an average spectator situated at a
distance equal to the width of the screen for viewing a “normal” scene. So as an example, the first pro-
ducers of television channels to broadcast in 3-D considered the size of an average television image to
be 42 inches with a viewing distance of about 3 meters.
A true proportionality between the offset of distant objects and the size of the screen is only exact for a screen having
Screen width
(base) % offset Offset for distant objects Correspondence in pixels
Screen of 26 m 0,25% 6,5 to 10 cm on the screen 5/7 pixels 2K or HD
Screen of 19,5 m 0,33 % 6,5 to 9 cm on the screen 7/8 pixels 2K or HD
Screen of 13 m 0,5% 6,5 to 8 cm on the screen 10/12 pixels 2K or HD
Screen of 9,85 m 0,66 % 6,5 to 7 cm on the screen 15/16 pixels 2K or HD

Sources : Alain Derobe-CST Janvier 2009
Screen of 6,5 m 1% 6,5 cm on the screen 20 pixels 2K or HD
Screen of 5 m 1,1 % 5,6 to 6 cm on the screen 22 pixels 2K or HD
Screen of 4 m 1,3 % 5 to 5,5 cm on the screen 27 pixels 2K or HD
Screen of 3m 1,6 % 4,8 to 5,3 cm on the screen 32 pixels 2K or HD
Screen of 2 m 2% 4 to 4,5 cm on the screen 40 pixels 2K or HD
Screen of 1,5 m 2,25 % 3,4 to 4 cm on the screen 45 pixels 2K or HD
Screen of 1 m 2,5 % 2,5 to 3 cm on the screen 50 pixels 2K or HD
Screen of 65 cm 2,3 % 1,5 to 3 cm on the screen 45 pixels 2K or HD
Screen of 50 cm 2,2 % 1 to 1,3 cm on the screen 42 pixels 2K or HD
Screen of 30 cm 2,25% 6 millimeters on the screen 40 pixels 2K or HD
Screen of 26 cm 2,3 % 5 millimeters on the screen 46 pixels 2K or HD
Screen of 13 cm 2,5% 3 millimeters on the screen 50 pixels 2K or HD

a base of around 6.5 meters. With larger screens, a bit larger offset of distant objects is possible, which will lead to a
divergence of only a fraction of a degree, even for the first rows of a movie theater. For a screen larger than 6.5 me-
ters at its base, another phenomenon comes into play: the synchrony that links convergence to the accommodation
of the eyes. This phenomenon systematically goes against the effort of parallelism, which allows seeing beyond the
surface of the screen, so that the distant offsets are limited to a maximum of 2.5% of the image’s width.

44 45

GLOSSARY 3Ds

3-D professional monitoring directly depends on the size of the screen it Accommodation: Modification of the curvature Disparity: Distance between the homologous
will be viewed on. of the lens of the eye due to the action of the points of the same object observed on the
Monitoring on a small size screen is useful
For example, on an editing monitor 1 me- muscles of the ciliary body in order to form images or the left and right retinas.
for validating errors in producing simple 3-D
effects such as inversion of the left and right ter wide with the viewer at a distance of 2 a clear image on the retina of objects obser- Focusing: An operation that consists of
meters, if the offset of the distant objects is ved at different distances. concentrating light rays coming from one
eyes, but it could never be used as a way of
appreciating 3-D in a feature film. adjusted from 40 to 50 pixels (2.5 %), the 3-D Angular disparity: See “Parallax angle”. point onto another point.
All the graphic artists and technicians in image could seem pleasant to watch, but re- Automultiscopy: 3Ds visualization without Ghosts: “Ghosting”, perception of “cross-
charge of an audiovisual and cinematogra- sult in a false setting according to the size of glasses with N viewpoints (N being greater talk” physical phenomena.
phic project must be able to appropriate the screen that the public will finally view in than 2). Homologous points: The corresponding points
stereoscopic production techniques and to a movie theater or a living room. It will then Autostereoscopy: 3Ds visualization without of the same object viewed by the left and ri-
regularly view rushes of their work, in par- be impossible for the images to be fused for glasses, with two viewpoints. ght eyes, and whose distance between the
ticular during editing to adjust the 3-D and the eyes of a viewer who looks at a 13-meter Binocular rivalry: The perception of diffe- eyes displayed on the retinas constitutes the
solve “depth continuity” problems, which screen (50 pixels instead of 10 pixels recom- rences between the left and right views, in retinal disparity. These corresponding points
results in retakes of the framing. What mended in the table above represents five particular with geometric and colorimetric of the same object are also present on the
would thus be ideal would be to have a 3-D times the ideal inter-ocular distance, that is asymmetry. left and right images displayed on the plane
screen available for each of the main pro- 32.5 cm on the screen). Convergence point: A point in space at the in- of the broadcast screen and whose distance
duction and postproduction stages in order However, it should be noted that the figures tersection of the optical axes of the eyes or between them constitutes the parallax.
to be able to validate the advancement of on the table represent an observed average cameras that converge towards it. Horizontal shifting: The horizontal movement
work in progress. It is also indispensable to and depend on the quality of the restitution Convergence: Rotation of both eyes (binocular of stereoscopic cameras made parallel du-
regularly test on “real-size” screens, that of the environment. Thus, a very good reso- convergence) or of two cameras, aiming at ring filming, or the displacement of two left
is, the same as those that viewers will be lution of a monitor allows the figures in the the same point in 3-D space. and right images in postproduction to set a
watching, for here also the impact of 3-D table to be slightly exceeded. Cristalline lens: Biconvex lens of the eye, null parallax and thus recompose the depth
behind the pupil, instrumental in the conver- of the scene.
gence of light rays towards the retina on Horopter A virtual arc of a circle of a person’s
which near or far observed objects are brou- visual environment formed by all the conver-
ght to a focus. gence points coming from the observation of
Crossed disparity: Retinal disparity corres- object-points located at the same distance
ponding to the crossing of the optical rays of from the eyes. All the points situated on this
each eye in front of the horopter or the plane curve correspond to identical places on each
of convergence. retina. The retinal disparity is thus zero on
Cross-talk: Imperfect physical dissociation of this curve. The disparity of objects observed
images respectively intended for the left and inside the horopter is called “crossed dispa-
right eyes during viewing of 3Ds content on a rity”, the disparity of objects situated outside
plano-stereoscopic screen. is called “non-crossed disparity”.
Depth range: The limits of the depth of 3Ds Hyperstereoscopy: Stereoscopy produced
content coming from the left and right image from a wide stereoscopic base, usually well
shots displayed on a screen or a stereosco- beyond the inter-ocular distance.
pic projection. Infinite stereoscopy: The greatest distance
Diopter A unit of homogenous vergence op- between the object and the viewer in bino-
posite the length. cular vision where the effect of depth is still
Diplopia: A disorder of vision characterized by discernible, usually estimated at 200 me-
the absence of fusion of the image perceived ters.
by each eye and provoking a double vision of Inter-axial distance: See “Stereoscopic base”.
the observed object. Inter-ocular distance: The distance between

46 47

the pupils of each eye, generally considered and right images), and one sees with diplopia Retinal disparity: Disparity between homolo- binocular vision. It allows the creation of left
for an adult male as being by default about vision. gous points of the same object formed on and right dual images and the perception of
65 mm. Parallax angle: The angle starting from the the retina of each eye. 3-D when broadcasting.
Keystone or trapezoid effect: The consequence of point of convergence observed by each eye Stereopsis: Sensation of depth given by bino- Vergence: The size characterizing the focu-
a shot with two cameras in convergence. and forming a triangle with them. cular vision. sing properties of a system.
Macrostereoscopy: Stereoscopy obtained from Parallax: The offset between the apparent Stereoscopic base: The distance between the Viewer: The observer, adult or child. Accor-
a small stereoscopic base, below the intra- positions of an object due to the distance optical axes of two lenses of a single camera ding to the difference in their respective in-
ocular distance, having virtually several mil- between the eyes of the observer. It is also or lenses of two cameras, with stereoscopic ter-ocular distances, adults and children do
limeters between each camera. the space between the left and right homolo- shooting. not have the same perception of 3-D depth
Magnification factor: The relation between the gous points of the same object observed on a Stereoscopic binocular acuity: The depth reso- and do not feel the same sensations of visual
rendered or captured image and the final size plano-stereoscopic screen. This space is due lution limit perceived by each eye. It repre- discomfort. It is thus necessary to take into
of this image on the broadcast screen. to the distance between the viewpoints of the sents the capacity to distinguish several account different distances to the screen
Maquette effect: A 3-D effect where objects observed scene. planes of depth close to each other, planes (children closer to a 3Ds screen than adults),
are perceived, during 3-D broadcasting, as Plane of convergence: When shooting in ste- defined by different parallax angles also a distance to be specified according to the
being smaller than in real life. This sensation reoscopy, the “depth” plane that is determi- close together. left and right view separation technology
is due to the chosen size of the stereoscopic ned by making the homologous points of a Stereoscopic fusion: A phenomenon that com- specific to each type of 3Ds screen, and to
base. The cameras being separated beyond scene’s object coincide, and where the came- bines in the brain the views coming from verify the inter-ocular distance hypotheses
the inter-ocular distance, the real world is ras have their optical axes in parallel. The po- each eye, allowing the perception of one tri- taken into account by each manufacturer.
perceived through the eyes of a giant and the sitioning is achieved by horizontal movement dimensional image. Zero parallax: A surface plane, physically re-
humans observed appear as ants. of the two cameras or by horizontal shifting in Stereoscopic window: The 3Ds region of the presented by the broadcast screen, on which
Motion control stereoscopy: Usage of the left postproduction. image in three dimensions represented by are situated all the left and right homolo-
and right cameras, especially to modify the Plano-stereoscopic screen: A screen from which the outer limits of the left and right views. gous points of the scene.
stereoscopic base, convergence, focusing, the perception of 3-D depth comes from the A horizontal movement between these left
zoom, time synchronization, calibration of display of viewpoints for the left and right and right views modifies the position of the
placement, and so on. eyes on exactly the same planar surface of stereoscopic window in space.
Negative parallax: 3-D objects perceived in front this screen. Stereoscopy: Principles and methods that
of the broadcast screen (for the retinal visual Plano-stereoscopic display: A display where 3-D allow the observation and/or restitution of
system, it is called “crossed disparity”. images intended for the left and right eyes
Non-crossed disparity: Retinal disparity cor- form on the same surface of a screen.
responding to the crossing of optical rays of Popping-out effect: When the stereoscopic win- APPENDIX 1: DEFINITION OF 3Ds PROFESSIONS
each eye in back of the horopter or the plane dow is shown on the plane of the screen, it

W
of convergence. has to do with the 3-D effect in which objects
ith the increase in 3-D experien- rections, the positioning of the plane of
Orthostereoscopy: The ideal position of the are seen in front of the physical broadcast
ce of teams on filming sites or in convergence according to the editing being
viewer in relation to the broadcast screen, screen. It corresponds to a negative parallax.
postproduction, as well as with the made, as well as the management of floa-
where the 3-D image of the filmed objects Positive parallax: 3-D objects perceived in back continuous training courses in stereogra- ting windows.
conforms to the real objects. The 3-D is per- of the broadcast screen (for the retinal visual phy for professionals, the stereographer It should be noted that the definition of 3-D
ceived without depth distortion (flattened or system, it is called “non-crossed disparity”). will be increasingly less necessary for trai- management positions are different accor-
stretched). Pseudoscopy: The inversion of the left and right ning everyone onsite. However, the stereo- ding to the genres (feature film, live broa-
Panum’s fusional area: A virtual region situated images respectively intended for the left and grapher will remain the one who ensures dcasts or TV programs). The methodology
around the horopter where the retinal dispa- right eyes. The left eye perceives the image the continuity of 3-D on a film or television and, above all, the teams will vary according
rities can be fused by the visual system into a intended for the right eye (and vice-versa). program in relation with the entire team to the budget allocated to the production.
tridimensional image. Outside of this region The 3-D is inversed and the objects normally working on a production. This is where the greatest difficulty lies: in
fusion is no longer possible, the images ap- situated in back of the scene are viewed in During postproduction, the stereographer making definitions of positions too defini-
pearing doubled (in reality, the non-fused left front, which causes visual discomfort. also ensures the follow-up on 3-D cor- tive. In spite of all that, we are attached to

48 49

this exercise in order to better understand duction stages such as correcting conver- For live shooting: gineer deals with 3-D vision and with supervi-
these new professions. gence errors, the continuity of depth in the sing the tools used for creating 3-D. He works
shots, compositing, calibrating, correction under the direction of the stereographer and
Stereographer is in charge of monitoring the angulation and
On stock TV and feature film programs: of disparities, etc.
On streaming TV programs being shot in the inter-axial distance through the remote
“Stereographer” or “Director of Stereography” 3Ds, the stereographer is usually employed control commands of 3-D assessment tools.
On a feature film, the stereographer has for Stereographic technician on the same level as the director or the head
When the means of production allow it, the He should not be confused with the traditio-
the most part the role of “stereo-supervisor” of photography. He conceives the general
stereographer can be assisted by a stereo- nal image vision engineer who ensures the
who determines with the director the “3-D stereographic options in collaboration with
graphic technician in charge of physical control of the image in agreement with the
shot” and the amplitude of the depth appro- the production and directing teams as well
adjustments of, or the 3-D systems used, head of photography by acting on the came-
priate for each sequence of the whole film. as the photography director and the chief ca-
in shooting. He thus takes the initiative in ras’ commands (diaphragm, level of blacks,
He also determines, with the photo direc- meraman (framing supervisor). He also par-
setting up the rig and the camera, the ver- whites and gammas, corrections of details
tor, the most appropriate hardware for the ticipates, as outside personnel, in conceiving
gence of the inter-axial distance and the and contours, matrices, etc.).
shoot. On productions having a large budget the filming of shots and in the positioning of
and high artistic ambitions, the stereogra- involuntary disparity problems. When there the camera. He then supervises the work
pher can be supported by a “stereographic are several teams shooting on a feature film of installing and adjusting the cameras and Stereographic technician
technician” dedicated to the operational ma- or several cameras on the same film set, the stereoscopic modules. He directs and super- In a large 3Ds multi-camera set-up (five to 6
nagement of the 3Ds settings and monito- different stereographic technicians can ap- vises the work of the “convergence-pullers” rigs), there is generally a stereographer who
ring as well as the 3Ds rushes. pear as generic “stereographers”, because (also sometimes called “assistant stereo- supervises all the filming, two 3-D vision
The stereographer must ensure the stereos- in this case they have much more autonomy graphers”) in order to obtain 3-D consistency engineers in the control room and a stereo-
copic “guarantee of good end results” of a than in a usual shooting environment. between cameras. He also assists the pro- graphic technician per active rig (or for seve-
production. The hierarchy remains unchan- duction in making choices that influence 3-D ral rigs). The vision engineer has a role very
ged, but the stereographer is the same as a 3-D postproduction supervisor effects. At the stereographer’s side is some- near to the one he has in 2-D, except that in
director of special effects under the director times found one or several stereographic addition he must pair (match) both cameras
and director of photography. The “3-D postproduction supervisor” belongs technicians who adjust the production rigs. on the same rig. It should be noted that tech-
- With the director, he studies the stereosco- to the laboratory team. He is responsible for Also found there are 3-D vision engineers in nologies evolve very quickly in terms of 3-D
pic variations for the entire film. the adjustment of the settings of the digital the control room. filming and the convergence settings, regu-
- With the chief operator, he studies the lat- postproduction machines. He will be working lated today by the stereographic technician,
ter’s demands to guarantee that the tools hand-in-hand with the stereographer as well could in the end become operations that are
3-D vision engineer more and more automated.
are used well and to respond to the specific as the editor and/or calibrator during 3-D ca-
Based in the control room, the 3-D vision en-
demands of the director of photography. He libration (depth grading). It also happens that
also participates, as outside personnel, in the stereographer intervenes in a film as a
how the scenes are shot and in how the ca- “3-D supervisor” from one end to the other APPENDIX 2: 3-D FILMING TECHNIQUES :
mera is positioned. He then supervises the of the production chain and thus takes on the From “Filming in Relief: the Complete and Interactive Course”, Parallell Cinema
work of installing and setting up the came- task of supervising the control of 3-D consis-
ras and stereoscopic modules. Stereoscopic offset its right image to the right. In this case, the

D
tency at different stages of postproduction. It
- For the production as a whole, he studies should be recalled that in postproduction it is e Stereoscopic offsets appear when object seems to be in the distance, behind
the necessary means for getting started in the image filmed by the right camera the plane of the screen, “in depth”.
necessary to correct both “visual disparities”
terms of teams and hardware. and the image filmed by the left ca- An offset is said to be null (0%) when the two
(keystone, alignment, optical aberrations,
- With all the production technicians, he mera are superimposed. The stereoscopic images of an object are perfectly superim-
etc.) and to bring “corrections to stereoscopic
answers their precise questions on adapting offset of an object, or parallax, is the dis- posed. In this case, the object appears on
settings” resulting from shooting errors, but
their professions to 3-D. tance to the image between left and right the plane of the screen (exactly at the depth
also from brutal 3-D variations introduced
- With the manager of postproduction, he images of an object. It is expressed in the of projection screen).
during editing by inverting shots planned for
studies the specific demands related to percentage of the width of the image, or in An offset is said to be negative (-1% for
in the beginning.
workflow. To ensure the consistency of all pixels: an offset is said to be positive when example) when the left image of an object is
the 3-D effects, he might supervise postpro- the left image of an object is to the left and to the right and its right image to the left. In

50 51

will often be horrible on a screen of 13 m.
They will give a weaker sensation of depth on
a screen of 2 or 3 m.
Adjusting the distant objects according to the
planned screen size is thus very important
from the beginning, even if it is possible to
make adjustments in postproduction.
Positive offset: the object seems Zero offset: the object seems on Zero offset: the object seems on • The artificial horizon of a 6.5 m screen is
behind the screen the plane of the screen the plane of the screen thus +1% (the “+” meaning a positive offset).
• The artificial horizon of a 13 m screen is
+0.5%.
• The artificial horizon of a 26 m screen is
+0.25%. Effect of “angulation” or “vergence” on the
The case where small screens are used is placement of the scenic box in relation to the
eye, the eyes will look straight in front of plane of the screen
them and be perfectly parallel in terms special: in fact, experience shows that the
of view. majority of viewers tolerate more easily an
On the other hand, if the two images on
the screen are now offset 13 cm, then
the left eye must look to the left and the
right eye to the right: the viewing enters
into divergence. Now this situation is not
natural for the human eye, and at best
it is tiring, or at worst painful, provoking
nausea and headaches.
It can thus be deduced that an offset (po-
Adjustment of Distant objects in sitive) on the screen of 6.5 cm between
distant objects divergence (diplopy) the two images of an object is the maxi-
on an artificial
horizon mum that we can tolerate.
For this reason, this offset value is of-
this case, the object seems near, in front of ten called the “artificial horizon”. It is
the plane of the screen, “popping out”. An naturally a question of an absolute va-
offset can be expressed, it should be noted, lue, which means that the same film is
in percentage or in number of pixels. not necessarily adapted to every screen
size. On a small screen (for example, 1
First step in shooting: adjusting distant objects m wide), if an offset is, for example, 1 cm
One of the basic rules to be respected during (that is, 1% of the width of the image),
shooting in 3-D is to ensure that the viewer it will be 13 cm when it is projected on
will really have the sensation that an object a screen of 13 m. It will be 26 cm on an
placed in the background of a scene is really IMAX© having a base of 26 m (see table
positioned in the “depth” of the screen. For on page 44: the relation between the 3Ds
example, the space existing between the parallax and the size of the screen).
eyes is on average 6.5 cm for an adult male. Therefore, distant objects offset +1% of
Therefore, if the image of a distant object the width of the image will be perfectly
intended for the left eye is offset 6.5 cm in adapted to a screen of 6.5 m (since our
relation to the image intended for the right +1% will make exactly 6.5 cm), but they
Effect of inter-axial distance (distance between the optical axes of the cameras) on the
stereoscopic range (amplitude of the scenic box)

52 53

artificial horizon situated between +1.8% and seems farther away. deformations of perspective in the angles of tant objects, the adjusting of the inter-axial
+2.5%, the latter value already being quite This method is used to adapt the 3-D image the image, called “keystoning”. These de- distance will also allow the position of the
large. Beyond this, the shot generally beco- to different screen sizes: an image is shifted formations, negligible in most shots, never- plane of the screen to be determined. There
mes too long to be superimposed. The artifi- in relation to another until distant objects theless require certain shots to be corrected exists several methods for determining the
cial horizon of a small screen is thus around correspond to the artificial horizon of each in postproduction using morphing effects to inter-axial distance necessary according to
+2% (with photos, the tolerance is greater be- screen size, that is: “make up for” the deformations of perspec- the scene being filmed.
cause the image does not move). • About +2% for small screens tive, which otherwise would create visual
It should be noted that with negative offsets • +1% for 6.5 m screens fatigue. Method for determining the inter-axial distance by
(that govern popping-out effects), no artificial • +0.5% for 13 m screens There exists two types of rigs: those that only
calculating the stereoscopic range
horizon really exists, convergence posing less • +0.25% for a 26 m IMAX© screen converge on one axis and those that conver-
If the farthest object in the image has an
problems than divergence. However, it must While shooting, it is useful to determine the ge on both so as to apportion the trapezoid
offset of +1% and the closest object has an
be kept in mind that an offset of -1% makes 3-D effect obtained by visualizing before- or “keystone” effect over both images.
offset of -2%, then the stereoscopic range of
an object pop out at ½ the viewer’s distance hand the effect of the offset by using a 3-D The convergence of the cameras defines the
this image is 3%, for +1-(-2)=3. Likewise, if
to the screen on a screen of 6.5 m, and at 2/3 monitor that allows an image to be shifted in plane of the screen and adjusts the distant
the farthest object has an offset of +2% and
this distance on one of 13 m, a very strong ef- relation to another. objects. It is necessary to locate the most
the nearest an offset of 0%, the stereoscopic
fect. distant object in the scene on a 3-D monitor
range of the image is 2%, for 2-0=2.
Filming with convergence and to angle a camera until both images of
For small screens, a stereoscopic range of
Filming in strict parallel t Another method exists for adjusting distant this object have the desired offset: for exam-
1.5% to 3% is often recommended. For lar-
Filming in strict parallel, without angulation objects, which is preferred by certain stereo- ple, +1%.
ger screens, smaller stereoscopic ranges
or offset of an image in relation to another graphers because it avoids the loss of a band are usually recommended: 0.75% to 1.5%.
one in postproduction, hardly takes place of several pixels on the left and right of the Second step: adjust the inter-axial distance Of course, the exact stereoscopic range
anymore except in IMAX© format, where it image when an image is shifted in relation to The inter-axial distance is often adjusted depends on each shot and the effect one
adapts well. For all smaller screens (cine- another in postproduction. Rather than shift once the distant objects have been set. The wishes to obtain. For certain shots (popping-
ma, television), this type of filming generally or offset an image in relation to another, a inter-axial distance is the distance separa- out effects, for example), one might want to
gives disappointing results, since it is cha- camera, or both, can be angled, that is, to ting the centers of each camera’s lens. This precisely control the stereoscopic range.
racterized by the absence of positive offsets: have them make a small panorama of less should not be confused with the inter-ocular Starting from an inter-axial distance of 0
everything is popping out, nothing is behind than one degree, which is also called the distance separating the optical axes of the cm, first the distant objects are set by using
the screen and distant objects are on the vergence of the cameras. eyes (6.5 cm on average). The inter-ocular angulation (for example, +1%). Then the in-
plane of the screen. When a camera is angled, a small panorama and inter-axial distances are independent, ter-axial distance is progressively increased
In strict parallel, without postproduction off- is made. Now, unlike a travelling, in a pano- the inter-axial distance being used to adjust until the left and right images of the closest
sets or angulation, as soon as a film is looked rama the perspective does not change. the “stereoscopic range” of an image, which object reach the value of the desired stereos-
at on a screen smaller than that of IMAX©, Mathematically, it is the same thing as a also depends on the focal length and the po- copic offset. This is directly controlled on the
popping out effects are not very comfortable postproduction offset (shift): with an angu- sitioning of the first and last object. 3Ds monitor (in our example, the inter-axial
due to the proximity of the screen, which for- lation, like in a postproduction offset, all the The inter-axial distance varies according to distance would be increased until a value of
ces the eyes to take quite large convergence pixels of the image are shifted to the left or the shots: when a large inter-axial distance -2% is reached for the closest object if a ste-
angles. to the right. This is why filming in parallel is needed (landscape shots, miniaturization reoscopic range of 3% is desired).
with postproduction offset and filming with effects), rigs are used side-by-side. But the- But another method can be used that has
Filming in parallel with offsets in postproduction convergence are mathematically equiva- se do not allow the cameras to get very close. the advantage of not needing a 3Ds monitor
lent—even if the two methods have practical When small inter-axial distances are needed thanks to the following formula:
(“shift”)
differences (keystone problems or loss of (for most shots), mirror rigs are used.
For this reason, on screens smaller than
pixels on the left and right of the image). As the inter-axial distance is used to adjust LxG=E
IMAX©, it is necessary to offset (“shift”) an
When filming with convergence, since one the stereoscopic range, it is set according to where:
image in relation to another one to create
camera is angled, in reality one camera the scene in order to obtain the stereoscopic L is the width (in meters) of the frame at the
positive offsets: this pushes all the 3-D in
films a rectangle and the other, by compari- range desired. But since the vergence of the level of the closest object (measured on the
a single block towards the back without
son, a trapezoid. Thus are introduced slight cameras was set to define the offset of dis- set, for example: 1.5 m).
touching the amount of depth; everything

54 55

G is the stereoscopic range desired in percent position of the plane of the screen is the most
(for example: 2%) important parameter. This allows, among mains beneficial for the operator, since a sin- Side-by-Side
E is the inter-axial distance in centimeters. other things, to avoid irritating window er- gle channel is enough to broadcast 2-D and
For example:1.5 x 2% = 3 cm rors. 3Ds images. But the necessity of a different
Once the distant objects are set through an- grammar makes editorial compatibility so-
gulation, the inter-axial distance is progres- mewhat delicate (axes and settings of the
Method for determining the inter-axial distance by shots, rhythm of editing, etc.). New produc-
sively increased until the plane of the screen
positioning the plane of the screen (that is, objects having an offset of 0%) is pla- tion methods need therefore to be invented
For other shots, where the stereoscopic ran- ced where one wants. in order to optimize the technical capabilities,
ge does not reach critical values, the precise all the while guaranteeing the artistic quality
of the 2-D and 3Ds versions of the same pro-
gram.
APPENDIX 3: CODECS AND DISTRIBUTION MODES

T
wo major technological families can be Frame Compatible modes of distribution
distinguished that allow stereoscopic The main principle of the Frame Compatible
modes is to reduce the resolution of the left The left and right views are made anamorphic
content to be broadcast. The first, called
and right images by half in order to group and placed side by side. This method induces
“Frame Compatible”, has the advantage that
them in one and the same HD image. This a theoretical horizontal loss of definition of
it can be used on existing broadcast networks.
spatial compression (horizontal or vertical 50% for each eye, but the work done by the
This technology spatially compresses the left
anamorphosis) generates a 2-D signal that brain to recreate the sensation of 3-D usually
and right views into a single image, but has
can be processed and transmitted by existing attenuates this loss of real definition.
the inconvenience of reducing by half the ori-
ginal resolution. Furthermore, a Frame Com- equipment.
patible image appears as two anamorphic During decoding, each of these two views is Top-and-Bottom
Horizontal decimation
images placed one beside the other, thus ma- (one pixel out of two is kept on each line) made “deanamorphic” and displayed on the
king the operator broadcast the 2-D version of screen in order to recreate the sensation of
the same program on a separate channel. 3-D. To compose a Frame Compatible image,
As for the “Service Compatible” mode, it alit is first necessary to select those pixels
lows one and the same signal to be broadcast to be retained in the left and right original
to all receivers. Using this type of approach, views. This step is called “decimation” and
2-D televisions only use that part of the signal there exists several ways of proceeding: one
that they need, but the 3Ds models are capa- can choose to only retain one pixel out of two
ble of displaying the left and right views in full on each line, in each column or in staggered
definition. rows. Moreover, the position of the retained
Whatever the method being considered, the pixels can be identical for both images, or
Vertical decimation
broadcast of a 3Ds signal requires a band- (one pixel out of two is kept in each column)
else offset (by one line or one column).
width much greater that its monoscopic Once the pixels to be retained have been se- The left and right views are made anamor-
equivalent. The Frame Compatible modes ge- lected in both views, then the way to arrange phic and placed one above the other. This
nerally require between 15% to 35% supple- them within one and the same image must method induces a theoretical vertical loss of
mentary throughput because of the increase be determined, which is called “packing”. definition of 50% for each eye, but, as for the
in high frequencies in the image that solicits Here again, several solutions exist: Side-by- Side-by-Side mode, the visual discomfort felt
the encoders more. Side, Top-and-Bottom, Checkerboard, etc. by the viewer is not necessarily very great.
As for the Service Compatible modes of distri- Although it is theoretically possible to choo-
bution, they require an even higher through- se different decimation packing modes, it is Line-by-Line
put, in the order of 50% to 70% in comparison generally understood that these two steps In this system, the lines coming from one
with the 2-D version. From a strictly technical Decimation in staggered rows
follow a certain logic. eye are placed on the even lines of the final
point of view, this increase of throughput re- (checkerboard sampling)

56 57

image and those coming from the other eye tical decimation.
on the odd lines. The “Line-by-Line” mode is However, the Checkerboard mode is not at
not very well adapted to video compression all adapted to video compression and will
because its structure solicits the encoders practically never be used such as it is for
more, and it works badly with color sub- broadcasting. On the other hand, decimation
sampling (4:2:0). by Checkerboard associated with Side-by- subtitle placed too near on a very deep bac- ther form of backward compatibility: it allows
Side packing has been shown to be a very kground. The standard allows this background full resolution 3Ds programs to be broadcast
Checkerboard efficient method for spatial compression to be managed differently according to the re- while remaining readable by a DVB-3D Phase
Each source image is sampled according to (Sensio proprietary codec, for example).at gion of the image where the subtitles become 1 decoder. The principle here is to transmit
a checkerboard structure and in a comple- the international level, but companies such as inserted. a Frame Compatible image accompanied by
Dolby and Sensio are already proposing pro- supplementary information that allows the
mentary way for each view. The loss of defi-
prietary methods. DVB-3DTV Phase 2 left and right views to be displayed in their full
nition felt is less than with horizontal or ver-
Les normes DVB-3DTV phase2 sont encore resolution. With this system, the first-genera-
eThe DVB-3DTV Phase 2 standards are still tion 3Ds decoders will process the 3-D images
being developed. They will permit a Service in half resolution per eye, whereas the new
DVB BROADCASTING IN 3Ds Compatible approach, that is, backward com- decoders will be capable of interpreting the
The standard describes the transmission of improvement layer and to reproduce the native
signaling information that allows the decoder patible with older television receivers:
- DVB-3DTV Phase 2a, whose specifications definition of the original images. The improve-
DVB-3DTV phase 1 to know which type of Frame Compatible it is ment layer transports, in fact, the spatial in-
The DVB European consortium, which propo- receiving, and thus to choose the appropriate should appear in the summer of 2012, will
allow TV viewers equipped with a 2-D screen formation not present in the base layer. Taking
ses TV broadcasting standards, defined 3Ds processing. This information is conveyed in the into account the strong correlation between
TV broadcasting modes in Frame Compatible SI (Service Information) tables of the DVB sys- to watch programs broadcast in 3Ds. This is
done by broadcasting the left image plus an these two streams of data, the improvement
format in February 2011. This broadcasting tem as well as in the MPEG elementary packets. layer proves to be very light (in the order of
can thus take place from now on using existing It has been planned that a channel can broad- improvement layer which will only be decoded
for 3-D screens. The 2-D televisions will not 10% of the base layer in current encoders).
HD infrastructures that make use of DVB stan- cast 2-D at certain moments and 3Ds at others: This approach has not yet been standardized
dards. in this case, the 2-D commutation to 3Ds and decode this and will only display the left eye.
The MVC codec can no doubt be used for this. at the international level, but companies such
This standard recommends H264 compression inversely can be automatically generated by as Dolby and Sensio are already proposing
as well as the use of the following display for- the decoder. Likewise, the electronic program Even so, this technique does not settle all pro-
duction problems: a production planned for proprietary methods.
mats: guide (EPG) will be able to say which programs
are in 3Ds. The standard foresees the dynamic 3Ds could turn out to be bad in 2-D.
management of subtitle and graphics paral- - DVB-3DTV Phase 2b corresponds to ano-
Side by side Top and Bottom
1080i/25 720p/50 lax so as to avoid conflicts of depth indications
1080i/29.97 720p/59.94 (a subtitle placed farther away than an object
1080i/30 720p/60 that it is hiding). This also allows visual fatigue
1080p/23.98 to be avoided caused by the juxtaposition of a

Professionnels du cinéma
et de l’audiovisuel, à vos
côtés tout au long de la vie
Retraite complémentaire
Assurance de personnes : prévoyance et santé
Action sociale et prévention
Médical
Services aux professions

© 2012 Ficam
Fédération des Industries du Cinéma, de l’Audiovisuel et du Multimédia
11/17, rue de l’Amiral Hamelin
75783 PARIS Cedex 16
Tél 01 45 05 72 55
Fax 01 45 05 72 50
Contact : Marc Bourhis : marc.bourhis@ficam.fr
_pour +
d’infos : Printing : Imprimerie Rimon
Adresse postale
11 Rue des Bintinais
ZA du Hil - 35230 Noyal Chatillon sur Seiche
02.99.53.85.54
0 d’un appel local 50 50
Prix
811 65 RCS SARL au capital de 300 000 EUROS
RCS Rennes B 329 778 435
SIRET 32977843500027
www.audiens.org Code APE : 1812Z

English version White Paper 3-D in Cinema and Television v2

English version White Paper 3-D in Cinema and Television v2

More Related Content

Similar to English version White Paper 3-D in Cinema and Television v2 (20)

More from Marc Bourhis (20)

English version White Paper 3-D in Cinema and Television v2