US007573475B2
`
`(12) Ulllted States Patent
`Sullivan et al.
`
`(10) Patent N0.:
`(45) Date of Patent:
`
`US 7,573,475 B2
`Aug. 11, 2009
`
`(54) 2D TO 3D IMAGE CONVERSION
`
`7,254,265 B2
`
`8/2007 Naske
`
`(75) I
`nventors:
`
`CA (Us)
`S F
`s H
`s
`teve u ivan, an rancisco,
`;
`Alan D- Trombla’ Fairfax’ CA (Us);
`Francesco G. Callari, San Francisco,
`CA (US)
`
`(73) Assignee: Industrial Light & Magic, San
`Francisco, CA (U S)
`
`7/2003 Naske et a1.
`2003/0128871 A1
`2004/0239670 A1 12/2004 Marks
`Zoos/0099414 Al
`5/2005 Kaye et 31'
`
`_
`(comlnued)
`
`( * ) Notice:
`
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(b) by 511 days.
`
`EP
`
`FOREIGN PATENT DOCUMENTS
`1 176 559
`V2002
`
`(21) App1.No.: 11/445,947
`
`(22) Filed:
`
`Jun. 1, 2006
`
`(65)
`
`Prior Publication Data
`
`US 2007/0279415 A1
`
`Dec. 6, 2007
`
`(51) Int. Cl.
`(2006.01)
`G06T 15/10
`(52) US. Cl. ..................... .. 345/427; 345/419; 345/426;
`345/581; 345/646; 345/647; 352/57
`(58) Field of Classi?cation Search ............... .. 345/419,
`345/426, 427, 581, 646, 647
`See application ?le for complete search history.
`
`(56)
`
`References Cited
`
`U.S. PATENT DOCUMENTS
`
`5/1979 Westell
`4,156,914 A
`4,972,359 A 11/1990 Silver et a1.
`5,613,048 A
`3/1997 Chen et a1.
`5,699,152 A * 12/1997 Fedor et a1. ............ .. 356/2401
`6,014,472 A
`1/2000 Minami
`6,208,348 B1
`3/2001 Kaye
`6,477,267 B1
`11/2002 Richards
`6,515,659 B1
`2/2003 Kaye et a1.
`6,686,926 B1
`2/2004 Kaye
`6,928,196 B1
`8/2005 Bradley et a1.
`7,102,633 B2
`9/2006 Kaye et a1.
`7,116,323 B2 10/2006 Kaye et a1.
`7,116,324 B2 10/2006 Kaye et a1.
`
`(Commued)
`OTHER PUBLICATIONS
`
`Bao et a1., “Non-linear View Interpolation,” J'. Wsualization Comp.
`Animation, 1999, l0:233_24l‘
`
`.
`(Commued)
`Primary ExamineriKimbinh T Nguyen
`(74) Attorney, Agent, or FirmiFish & Richardson PC.
`
`(57)
`
`ABSTRACT
`
`A method of creating a complementary stereoscopic image
`pair is described. The method includes receiving a ?rst 2D
`image comprising image data, Where the ?rst 2D image is
`captured from a ?rst camera location. The method also
`includes projecting at least a portion of the ?rst 2D image onto
`computer-generated geometry. The image data has depth val
`ues associated With the computer-generated geometry. The
`system includes rendering, using the computer-generated
`geometry and a second camera location that differs from the
`?rst camera location, a second 2-D image that is stereoscopi
`cally complementary to the ?rst 2-D image, and in?lling
`image data that is absent from the second 2-D image.
`
`38 Claims, 13 Drawing Sheets
`
`PRIME FOCUS EX 1006-1
`PRIME FOCUS v LEGEND3D
`
`

`
`US 7,573,475 B2
`Page 2
`
`US. PATENT DOCUMENTS
`
`2005/0104878 A1
`2005/0104879 A1
`2005/0117215 A1*
`2005/0146521 A1
`2005/0231505 A1
`
`5/ 2005 Kaye et al.
`5/ 2005 Kaye et al.
`6/2005 Lange ...................... .. 359/462
`7/2005 Kaye et al.
`10/2005 Kaye et al.
`
`FOREIGN PATENT DOCUMENTS
`
`20070042989
`KR
`WO WO/2005084298
`W0 WO 2006/078237
`WO WO/2006078249
`WO WO/2006078250
`
`4/2007
`9/2005
`7/2006
`7/2006
`7/2006
`
`OTHER PUBLICATIONS
`
`Bao et al., “Non-linear View Interpolation,” Computer Graphics and
`Applications, Paci?c Graphics ’98 Sixth Paci?c Conference, 1998,
`pp. 61-69, 225.
`Chen and Williams, “View Interpolation for Image Synthesis,” Proc.
`Of the 20th Annual Conference on Computer Graphics and Interac
`tive Techniques, 1993, 279-288.
`Fu et al., “An Accelerated Rendering Algorithm for Stereoscopic
`Display,” Comp. & Graphics, 1996, 20(2):223-229.
`
`McMillan and Bishop, “Head-tracked stereoscopic display using
`image Warping,” Stereoscopic Displays and Wrtual Reality Systems
`11, Fisher et al. (eds.), SPIE Proceedings 2409, San Jose, CA, Feb.
`5-10, 1995, pp. 21-30.
`Raskar, “Projectors: Advanced Graphics and Vision Techniques,”
`SIGGRAPH 2004 Course 22 Notes, 166 pages.
`SaWhney et al., “Hybrid Stereo Camera: An IBR Approach for Syn
`thesis of Very High Resolution Stereoscopic Image Sequences,”
`ACM SIGGRAPH 2001, Aug. 12-17, 2001, Los Angeles, CA, USA,
`pp. 45 1 -460.
`DeskoWitZ, Chicken Little Goes 3-D With Help from ILM, “VFX
`World”, (Nov. 7, 2005) 2 pages.
`U.S. Appl. No. 11/446,576, ?led Jun. 1, 2006, Non-?nal Of?ce
`Action, mailed Feb. 27, 2008, 26 pages.
`U.S. Appl. No. 11/446,576, ?led Jun. 1, 2006, response to Feb. 27,
`2008 Non-?nal Of?ce Action, 14 pages.
`U.S. Appl. No. 11/446,576, ?led Jun. 1, 2006, Non-?nal Of?ce
`Action, mailed Sep. 12, 2008, 10 pages.
`U.S. Appl. No. 11/446,576, ?led Jun. 1, 2006, response to Sep. 12,
`2008 Non-?nal Of?ce Action, 13 pages.
`U.S. Appl. No. 11/446,576, ?led Jun. 1, 2006, Final Of?ce Action,
`mailed Nov. 26, 2008, 11 pages.
`U.S. Appl. No. 11/446,576, ?led Jun. 1, 2006, response to Nov. 26,
`2008 Final Of?ce Action, 14 pages.
`* cited by examiner
`
`PRIME FOCUS EX 1006-2
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 1 0f 13
`
`US 7,573,475 B2
`
`fllIk
`
`
`
`COSWELOFE QOWF: Om
`
`@mEmO
`29m
`
`>65
`
`3
`
`#2.
`
`92
`
`glw
`
`>>o_>
`
`mLmEmo :3
`
`
`
`COSNELOFE QOQF: ON
`
`mmmhouw
`
`PRIME FOCUS EX 1006-3
`PRIME FOCUS v LEGEND3D
`
`

`
`PRIME FOCUS EX 1006-4
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 3 0f 13
`
`US 7,573,475 B2
`
`3001
`
`Build left camera
`View
`£72
`1
`
`_ .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`.
`
`.
`
`. .'
`
`; Build geometry for ;
`;
`left camera view 3
`M I
`1 .... .1 ..... ..
`
`Project left view
`on geometry
`3'26
`
`i Calculate left and
`
`right depth maps 4___—_
`M
`i
`Calculate offset
`map (see FlG. 10)
`£0
`
`Combine right
`camera view and
`left camera view
`
`information to
`
`create 30 Film
`information
`if
`I
`
`Run compositing
`script to generate
`n'ght camera view
`information
`
`T
`
`Create initial right
`camera images/
`Execute in?ll
`algon'thm
`
`M
`
`ls pixel
`part of 2D scene
`element?
`
`LL?
`
`Yes
`
`ls pixel
`occluded?
`
`ilé
`
`'
`Execute blurring
`algorithm
`
`3.11
`
`I ,
`
`FIG. 3
`
`PRIME FOCUS EX 1006-5
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 4 0f 13
`
`US 7,573,475 B2
`
`55:.
`
`.8»
`A1
`
`L1
`E»
`
`1N a> ex.
`
`EN
`
`PRIME FOCUS EX 1006-6
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 5 0f 13
`
`US 7,573,475 B2
`
`l
`
`Start
`
`‘F
`
`End
`
`500
`
`Compute 3D point -
`at distance from
`left camera center
`£213
`+
`Compute 20 point
`in ?eld of view of
`right camera
`M
`
`ifferenc
`between the
`expected distance
`and the measured
`distance exceed a
`predetermined
`threshold?
`
`Set left-right offset
`map for pixel as
`“background’
`M9
`
`Yes
`
`Distance
`from the left
`camera view to the pixel >=
`the background
`threshold?
`_5_0_4
`
`'
`
`Set left-right offset
`map for pixel as
`"outside"
`i7
`
`20
`point
`outside ?eld
`
`Set the distance of the
`2D point as the
`distance from the right
`camera center
`i6
`+
`Set coordinates of 20
`point as the integer
`closest to the 20 point
`if
`
`Distance
`from the camera
`view to the pixel exceed
`background
`threshold?
`M7
`
`FIG. 5
`
`‘
`
`Set left-right offset
`map for pixel as
`“occluded”
`.521
`
`Yes
`
`Set left-right offset
`map for pixel to a ‘
`distance value
`if
`
`PRIME FOCUS EX 1006-7
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 6 0f 13
`
`US 7,573,475 B2
`
`MSW
`
`PRIME FOCUS EX 1006-8
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 7 0f 13
`
`US 7,573,475 B2
`
`K
`\
`I / I I /
`\ \ \ \.
`\
`/
`\ I‘
`/ ll
`
`\|\r|l)
`
`
`
`Ema U=m> o2
`\l
`%Q..\
`
`mn .OE
`
`Ema 26> o2
`XI
`
`(h .9“
`
`PRIME FOCUS EX 1006-9
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 8 0f 13
`
`US 7,573,475 B2
`
`PRIME FOCUS EX 1006-10
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 9 0f 13
`
`US 7,573,475 B2
`
`(
`
`Start
`
`)
`
`V
`Receive next pixel
`of light camera ‘
`view
`L2
`
`is offset
`value of pixel
`‘background’?
`
`is offset
`value of pixel
`‘occluded’ or ‘outside’?
`
`Yes
`
`900K
`
`Assign right camera pixel
`color information from a
`pixel of the left view
`location at same 2D
`coordinates
`M
`
`Assign right camera pixel a
`value of “occluded“
`
`Assign right camera pixel a
`pixel color of a left camera
`pixel nearest to a location
`speci?ed by the 2D
`coordinates of the right
`camera pixel plus the offset
`value
`
`FIG. 9A
`
`Yes
`
`More
`pixels?
`M
`
`No
`
`PRIME FOCUS EX 1006-11
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 10 0f 13
`
`US 7,573,475 B2
`
`9502
`
`Yes
`
`More
`pixels?
`w
`
`No
`
`No
`
`(
`
`Start
`
`>
`
`Y
`Receive next pixel of
`rig ht camera view
`25.2
`
`4
`
`is offset
`value of pixel
`‘occluded’?
`2L4
`
`Yes
`
`Locate the closest
`unoccluded pixel to the
`le? of the occluded pixel
`gig
`
`Does
`the unocclude
`pixel exist?
`
`Yes
`
`Apply the
`noccluded pixel
`U
`color to the
`occluded pixel
`m
`A
`
`Locate the closest
`unoccluded pixel to the
`right of the occluded
`pixel
`ii?
`
`Does
`the unoccluded
`pixel exist?
`M
`
`No
`
`Yes
`
`Report Error
`26_6
`
`FIG. 98
`
`End
`
`PRIME FOCUS EX 1006-12
`PRIME FOCUS v LEGEND3D
`
`

`
`U.S. Patent
`
`Aug. 11,2009
`
`Sheet 11 0f 13
`
`US 7,573,475 B2
`
`11006
`
`1 1002
`
`A
`
`10001
`
`
`
`FIG. 105
`
`FIG. 10A
`
`W
`
`216-;
`
`PRIME FOCUS EX 1006-13
`PRIME FOCUS v LEGEND3D
`
`

`
`US. Patent
`
`Aug. 11,2009
`
`Sheet 12 0f 13
`
`US 7,573,475 B2
`
`Use le? eye camera view
`and uncorrected right
`camera view for next
`frame
`110.?
`
`l
`
`De?ne keystone
`correction plane (P) for left
`camera view
`1104
`
`1
`
`l
`
`Compute line L, which is
`the intersection of P with
`uncorrected right camera
`view
`1101?
`
`Compute left most point
`(A) of intersection L with
`the right camera view
`frustum
`1108
`
`1
`
`Compute point B of
`intersection of P with line
`from the optical center (C)
`of the right camera
`parallel to the optical axis
`of the left camera view
`1&0
`
`Yes
`
`More
`Frames?
`1122
`
`Linearly warp the offset
`using homography
`1120
`
`l
`
`Replace uncorrected right
`camera view with virtual
`right camera view
`1118
`
`l
`
`Compute homography
`between image planes
`1116
`
`T
`
`De?ne a “virtual” right
`camera view using f, C
`and a parallel movement
`relative to the left camera
`view
`iii
`
`l
`
`Compute the width (f) of
`the angle ACB
`1112
`
`j
`
`FIG. 11
`
`PRIME FOCUS EX 1006-14
`PRIME FOCUS v LEGEND3D
`
`

`
`PRIME FOCUS EX 1006-15
`PRIME FOCUS v LEGEND3D
`
`

`
`US 7,573,475 B2
`
`1
`2D TO 3D IMAGE CONVERSION
`
`CROSS REFERENCE TO RELATED CASES
`
`This application is related to US. patent application Ser.
`No. 11/446,576 ?led Jun. 1, 2006 by Davidson et al., and
`entitled “In?lling for 2D to 3D Image Conversion,” the
`entirety of Which is incorporated herein by reference.
`
`TECHNICAL FIELD
`
`This speci?cation relates to systems and methods for 2D to
`3D image conversion.
`
`BACKGROUND
`
`Moviegoers can vieW specially formatted ?lms in three
`dimensions (3D), Where the objects in the ?lm appear to
`project from or recede into the movie screen. Formatting the
`3D ?lms can include combining a stereo pair of left and right
`images that are tWo-dimensional. When the stereo pair of
`images is combined, a vieWer Wearing polariZed glasses per
`ceives the resulting image as three-dimensional.
`
`SUMMARY
`
`The present speci?cation relates to systems and methods
`for converting a 2D image into a stereoscopic image pair.
`In a ?rst general aspect a method of creating a complemen
`tary stereoscopic image pair is described. The method
`includes receiving a ?rst 2D image comprising image data,
`Where the ?rst 2D image is captured from a ?rst camera
`location. The method also includes projecting at least a por
`tion of the ?rst 2D image onto computer-generated geometry.
`The image data has depth values associated With the com
`puter-generated geometry. The system includes rendering,
`using the computer- generated geometry and a second camera
`location that differs from the ?rst camera location, a second
`2-D image that is stereoscopically complementary to the ?rst
`2-D image, and in?lling image data that is absent from the
`second 2-D image.
`In a second general aspect, a method of identifying an
`absence of image information When generating a comple
`mentary stereoscopic image pair is described. The method
`includes illuminating at least a portion of computer- generated
`geometry using a simulated light source positioned at a ?rst
`camera location, designating a second camera at a second
`camera location having a vieWing frustum that includes the at
`least a portion of the computer- generated geometry, and iden
`tifying an absence of valid image information at locations
`having shadoWs produced by the simulated light When the
`locations are vieWed from the second camera.
`In a third general aspect, a system for identifying an
`absence of valid pixel data in a stereoscopically complemen
`tary image is described. The system includes an interface to
`receive a ?rst 2D image captured from a ?rst camera position,
`a projector to project a portion of the ?rst 2D image onto
`computer-generated geometry that corresponds to the por
`tion, a rendering module to render from a second camera
`position a second 2D image that is stereoscopically comple
`mentary to the ?rst 2D image using the projected portion and
`the computer-generated geometry, and a depth calculator to
`determine depths of image information in the ?rst and second
`images for use in identifying an absence of valid image infor
`mation in the second 2D image.
`In another general aspect, a computer program product
`tangibly embodied in an information carrier is described. The
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`2
`computer program product includes instructions that, When
`executed, perform operations including aligning computer
`generated geometry With a corresponding portion of a ?rst 2D
`image captured from a ?rst camera location, projecting the
`portion of the ?rst 2D image onto the geometry, designating a
`second camera location that differs from the ?rst camera
`location, using the geometry and the designated second cam
`era location to render a second 2D image that is stereoscopi
`cally complementary to the ?rst 2D image, and identifying an
`absence of valid image information in the second 2D image
`by comparing depth values of image information from the
`?rst and second 2D images.
`In yet another general aspect, a method of identifying an
`absence of image information When generating a comple
`mentary stereoscopic image pair is described. The method
`includes projecting at least a portion of a ?rst 2D image
`including image information onto computer- generated geom
`etry from a ?rst camera having a ?rst vieWing frustum at a ?rst
`camera location, designating a second camera having a sec
`ond vieWing frustum at a second camera location. The second
`vieWing frustum includes at least a portion of the computer
`generated geometry. The method also includes determining
`depth values associated With the image information in the ?rst
`vieWing frustum and in the second vieWing frustum, Wherein
`each of the depth values indicates a distance betWeen one of
`the cameras and a point in the projected ?rst 2D image in the
`vieWing frustum for the associated camera, and identifying an
`absence of valid image information by comparing the depth
`values for corresponding points of the projected ?rst 2D
`image in the ?rst and second vieWing frustums.
`The systems and techniques described here may provide
`one or more of the following advantages. First, a system may
`effectively associate image information elements With depth
`values that lack a description of their three-dimensional
`geometry or position, Which can increase the ?exibility of the
`system and the speed of 2D-to-3D conversion processes.
`Second, a system can produce a visually compelling stereo
`scopic image pair even When the focal axes of the stereo
`cameras are not parallel. Third, a system can produce a visu
`ally compelling second vieW using unoccluded portions of an
`image that are associated With other portions of the image that
`are geometrically occluded When vieWed from a ?rst camera.
`The details of one or more embodiments of systems and
`methods for 2D to 3D image conversion are set forth in the
`accompanying draWings and the description beloW. Other
`features and advantages of the embodiments Will be apparent
`from the description and draWings, and from the claims.
`
`DESCRIPTION OF DRAWINGS
`
`FIG. 1 is a schematic diagram of an example of a system for
`generating 3D image information.
`FIG. 2 is a block diagram of an example of a 2D-to-3D
`converter shoWn in the system of FIG. 1.
`FIG. 3 is a How chart illustrating a method for creating 3D
`image information.
`FIGS. 4A and 4B are schematic diagrams illustrating the
`generation of an offset map in a left camera vieW and a right
`camera vieW, respectively.
`FIG. 5 is a How chart illustrating a method for generating an
`offset map.
`FIG. 6A is a schematic diagram illustrating a perspective
`vieW of a difference betWeen a presence of data in a left
`camera vieW and a right camera vieW.
`FIG. 6B is a schematic diagram illustrating a top doWn
`vieW of a difference betWeen a presence of data in a left
`camera vieW and a right camera vieW.
`
`PRIME FOCUS EX 1006-16
`PRIME FOCUS v LEGEND3D
`
`

`
`US 7,573,475 B2
`
`3
`FIGS. 7A and 7B are block diagrams illustrating tWo
`approaches for ?lling gaps in pixel data.
`FIG. 8A is a block diagram illustrating a generation of
`pixel data in a right camera vieW Where a left camera vieW
`contained no prior valid pixel data.
`FIG. 8B is a block diagram illustrating an exploded vieW of
`the generation of pixel data shoWn in FIG. 8A.
`FIGS. 9A and 9B shoW methods for mapping pixel values
`to pixels included a right camera vieW.
`FIG. 10 is a schematic diagram illustrating a difference in
`camera orientation.
`FIG. 11 is a How chart illustrating a method for correcting
`differences betWeen camera orientations.
`FIG. 12 is a block diagram of a general computer system.
`Like reference symbols in the various draWings indicate
`like elements.
`
`DETAILED DESCRIPTION
`
`FIG. 1 illustrates a system 100 for converting 2D image
`information 102 to a stereoscopic pair of images that may be
`perceived by a vieWer as three-dimensional. The stereoscopic
`pair of images is referred to as 3D image information 104. The
`system 100 can generate a stereoscopic image pair by align
`ing computer-generated geometry that corresponds to objects
`shoWn in a ?rst 2D image included in the 2D image informa
`tion 102. The ?rst 2D image canbe projected on the geometry,
`and a second camera that is offset from the ?rst camera can
`render a second 2D image that is complementary to the ?rst
`2D image. Image data projected from the ?rst camera may not
`be visible from the second camera (e. g., due to occlusion the
`image data by geometry), can be generated using in?lling
`techniques described in more detail beloW. The ?rst 2D image
`and the second 2D image can be vieWed together as a stereo
`scopic image pair that can be consistent With an actual 3D
`scene.
`In certain embodiments, the 2D image information 102
`includes a ?rst camera vieW, such as a left camera vieW 106,
`Which is used to create a second camera vieW, such as a right
`camera vieW 108. Combining the left camera vieW 106 and
`the right camera vieW 108 can produce stereoscopic 3D image
`information 104. A 2D-to-3D converter 110 can receive 2D
`information 102 from a variety of sources, for example, exter
`nal or removable storage 112, optical media 114, another
`Workstation 116, or a combination thereof.
`The 2D-to-3D converter 110 can use the 2D information
`102 to generate the right camera vieW 108. In certain embodi
`ments, generation of the right camera vieW 108 can result in
`areas that lack valid data. In?lling, Which is illustrated in FIG.
`7 and described in more detail in FIG. 8, can use valid pixel
`data from surrounding areas to ?ll in the areas that lack valid
`data.
`The 2D-to-3D converter 110 can combine the left camera
`vieW 106 and the right camera 108 vieW to create 3D image
`information 104. The 3D image information 104 can be used
`to generate stereoscopic images that appear in three dimen
`sions When vieWed With stereoscopic glasses, such as Real
`D’ s Cinema glasses, manufactured by Real D, Inc. of Beverly
`Hills, Calif.
`FIG. 2 illustrates an example 200 of a 2D-to-3D converter
`110, Which receives 2D image information 102 and generates
`3D image information 104. The 2D image information 102,
`shoWn in FIG. 2, can include left camera information 202, any
`number of ?nal images 204, render elements 206, non-ren
`dered elements 212, and optionally, a compositing script 210,
`such as a script generated by Shake® developed by Apple
`Computer Inc. (Cupertino, Calif.). The 2D image information
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`4
`102 may also include or be supplemented With three-dimen
`sional information, for example, the fundamental geometry
`208. The 2D-to-3D converter 110 includes a geometry ren
`derer 214, a left camera 216 that includes a projector 218, a
`depth calculator 220, a right camera 222 that includes a ren
`dering module 224, and a compositing script module 226.
`These subsystems 214-226 can be used by the 2D-to-3D
`converter 110 to generate a scene that includes 3D image
`information 104. Each 2D scene received by the 2D-to-3D
`converter 110 can include ?nal images 204. Final images can
`include visual components of an image scene, for example, a
`protagonist or antagonist of a ?lm, and the surrounding envi
`ronment, such as buildings, tress, cars, etc. Each ?nal image
`204 can be associated With pieces of fundamental geometry
`208. The fundamental geometry 208 can de?ne the location
`of a computer-generated object in three-space and can include
`one or more rendered elements 206.
`Each rendered element can include color and texture infor
`mation. For example, a ?nal image of a car can be associated
`With fundamental geometry describing the shape of the car.
`The fundamental geometry of the car can include rendered
`elements, such as a cube for the body and spheres for the
`Wheels. The rendered body element can include properties,
`for example a metallic red texture, and the rendered Wheel
`elements can include properties, such as a black matte rubber
`like texture.
`Final images can also be associated With non-rendered
`elements 212 that are not associated With fundamental geom
`etry (e.g., hair, or smoke), and thus may not be de?ned in
`three-dimensional space.
`The compositing script module 210 can use the ?nal
`images 204, the fundamental geometry 208, and non-ren
`dered elements 212 to generate timing information used to
`correctly composite a scene. Timing information can specify
`the rate at Which different objects move Within a scene and the
`obj ect’s position at various times. The systems and tech
`niques described in this speci?cation can improve the consis
`tency of the second, or right image, When vieWed over a series
`of sequential frames. This can decrease the appearance of
`jitter, or noise, betWeen the frames When the series of frames
`is vieWed as a ?lm.
`For example, a scene can include a person Walking to the
`left of the scene, a person Walking to the right of a scene, and
`an object falling from the top of scene. If a scene designer
`Wants both individuals simultaneously to arrive at the center
`of a frame so that the falling object hits them, the compositing
`script module 210 uses timing information included in the
`compositing script to coordinate the necessary placement of
`objects. If the object falls too quickly, the timing information
`associated With the falling object may be changed to sloW the
`descent of the object. The rate at Which either the person
`Walking to the right or the person Walking to the left can also
`be modi?ed so that they meet at the same time the falling
`object reaches them.
`In certain embodiments, a compositing script module 210,
`such as Shake® is used. The compositing script module 210
`can use a compositing script that can include a set of opera
`tions that, When executed, create the composited scene. The
`operations can include color correction, keying, and various
`image transform operations. Color correction is an operation
`that can alloW the artist to modify the color of a scene, for
`instance adding more blue to a sky scene. Keying is a com
`positing technique that can combine tWo images, for instance,
`the sky image can be combined With an image of a car in front
`of a “blue screen” or “green screen” to create an image of the
`car in front of the sky image. Image operations can include
`image rotation, image scaling, and image repositioning. For
`
`PRIME FOCUS EX 1006-17
`PRIME FOCUS v LEGEND3D
`
`

`
`US 7,573,475 B2
`
`5
`instance, an image transform operation can scale the keyed
`image of the car and sky to increase its siZe to tWo times its
`original siZe, or in another example, the image transform
`operation can rotate the keyed image 180 degrees such that
`the image is turned upside doWn.
`Operations can be vieWed in context, alloWing the end
`results of an operation to be vieWed While the user con?gures
`the properties of the operation. The compositing script mod
`ule 210 may permit the ?nal images 204 to be unaffected by
`changes in scene timing. For example, a scene consisting of
`tWo cars, Where the ?rst car is supposed to jump off a ramp,
`and the second car is supposed to simultaneously drive under
`neath the jumping car. In certain timing structures, the jump
`ing car may reach the ramp too quickly or too sloWly, throW
`ing off the desired effect of the scene. In cases Where the scene
`timing is incorrect, the compositing script may be changed
`using the neW timing information, and the ?nal images 204
`may not need to be re-rendered. Timing changes can come
`from a variety of sources, for example, at a request by a ?lm
`director.
`FIG. 3 illustrates a method 300 for creating 3D image
`information 104. The 2D-to-3D converter 110 can generate a
`left camera, as shoWn by step 302. In certain embodiments,
`this includes receiving the left camera information 202, such
`as the camera’s aspect ratio, ?eld of vieW, vieW vector, and
`position. The 2D-to-3D converter 110 can create the left
`camera 216 With the left camera information 202.
`The 2D-to-3D converter 110 may optionally build funda
`mental geometry 208 for the camera vieW, as shoWn in step
`304. In certain embodiments, hoWever, the fundamental
`geometry 208 can be received by the 2D-to-3D converter 110,
`and thus does not need to be built. The projector 218 can
`project the left camera vieW’s ?nal images 204 onto the left
`camera vieW’s fundamental geometry 208, as shoWn in step
`306.
`A right camera 222 can be generated by the 2D-to-3D 110
`converter and positioned at a location offset relative to the left
`camera 216. In certain embodiments, this distance is substan
`tially similar to the average interocular distance betWeen the
`human eyes, Which is approximately 6.5 cm. In other embodi
`ments, this distance can be set by an artist to achieve a desired
`3D effect, such as 1/3oth of the distance betWeen the camera
`and nearest object in the scene.
`The right camera 222 can use the rendering module 224 to
`render the 2D image information’s fundamental geometry
`208 using the right camera 222 information. Rendering can
`de?ne an object through various techniques that measure the
`outgoing light from the surface of an object. In certain
`embodiments, the rendering module 224 can render using a
`ray tracing technique that casts a ray to each pixel in the scene
`to generate an image. In other embodiments, the rendering
`module 224 can render using a primitive-by-primitive
`method, Where the object is de?ned as a set of primitives (e.g.,
`triangles, polygons, etc.) and each primitive is then rasteriZed,
`Which converts the vector data of the primitive into a form that
`can be displayed. For example, using vector information that
`de?nes a cube and generating red, green and blue (RGB)
`color values that are displayed on a monitor.
`In certain embodiments, a depth map can be created, as
`shoWn in step 308, to help identify pixels that may be
`occluded or missing When the stereoscopically complemen
`tary image (e.g., the right camera vieW 108) is created based
`on a given image (e.g., the left camera vieW 106). The depth
`maps can be created using rasteriZation and depth-buffering
`algorithms. A depth map can be a tWo-dimensional array that
`has the same resolution as the current image scene. For
`
`20
`
`25
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`6
`example, for each pixel located at (X1, Y1) in the scene, there
`exists a location in the depth map referenced by the same Ci 1,
`Y1) coordinates.
`Each location of the depth map can contain depth informa
`tion for a pixel in the scene. This depth information can also
`be thought of as the pixel’s distance coordinate in a three
`dimensional polar coordinate system. For example, after ref
`erencing the depth map, a 2D pixel at Gil, Y1) can use the
`depth value Z l retrieved from the depth map indexed at Q( 1,
`Y1) to de?ne a 3D ray oriented from the projection center of
`the camera to the pixel, and a 3D point at distance Z 1 from the
`optical center along the positive direction of that line. In some
`embodiments, the depth information may contain more than
`one depth value for a pixel. For example, if a transparent
`object includes a pixel, that pixel may be associated With
`multiple depthsione depth for images vieWed through the
`object from one angle and another depth for images vieWed
`through the object at another angle.
`The depth calculator 220 can obtain the pixel depth value
`of a pixel in the left camera vieW 106 or the right camera vieW
`108 by referencing the tWo-dimensional depth map for the
`pixel in either the left camera vieW 106 or the right camera
`vieW 108, respectively.
`The tWo sets of depths values, from the left camera vieW
`106 depth map and the right camera vieW 108 depth map, can
`be used to identify a set of corresponding pixels from each
`camera vieW Where the pixels from one camera vieW corre
`spond to pixels from the alternate camera vieW. In some
`embodiments, the identi?cation proceeds as described in
`FIG. 5. Speci?cally, for each pixel of the left camera vieW
`Whose distance coordinate is less than a pre-de?ned “back
`groun ” threshold, a 3D point location is computed at step
`508. A corresponding pixel location in the right camera vieW
`is computed at step 510, and compared With the right-camera
`?eld of vieW at step 512. If it is inside that ?eld of vieW, and
`its distance from the right camera center is less than a pre
`de?ned “background” threshold, then the distance itself is
`compared at step 522 With the distance recorded at the closest
`2D pixel location in the depth map associated With the right
`camera. If the tWo distances are Within a pre-determined
`“occlusion” threshold, the tWo pixel locations are said to be
`corresponding, otherWise both pixel locations are said to be
`“occluded”.
`After corresponding pixels are identi?ed, the 2D-to-3D
`converter 110 can calculate an offset map, as shoWn in step
`310. The offset map can be calculated by determining the
`change in position betWeen a pixel in the left camera vieW 1 06
`and the corresponding pixel in the right camera vieW 108. For
`example, if the coordinates of point P in the left camera vieW
`106 are de?ned at (i, j), Where point P is not occluded, and the
`right camera vieW 108 de?nes the coordinates of the point
`corresponding to P as (x, y), the offset map may contain the
`2D vector o(x-j, y-i). This process is further explained in
`associated With FIG. 5.
`In certain instances, pixels in the left camera vieW 106 may
`not share similar depth values With pixels in the right camera
`vieW 108. If the pixels do not share similar depth values, it
`may indicate the valid pixel data is missing or occluded. In
`certain embodiments the 2D-to-3D converter 110 can execute
`a blurring algorithm to modify the offset map.
`In certain embodiments, the offset maps do not contain any
`valid depth information for the non-rendered elements 212,
`such as smoke. HoWever, the depth of the non-rendered ele
`ments 212 may be similar to the depth of nearby sections of
`fundamental geometry 208 or rendered elements 206. For
`
`PRIME FOCUS EX 1006-18
`PRIME FOCUS v LEGEND3D
`
`

`
`US 7,573,475 B2
`
`7
`example, the depth of a feather may be similar to the depth of
`a head if the feather is near the head in the left camera vieW’ s
`?nal image 204.
`If a pixel in the left camera vieW 106 is part of a non
`rendered 2D scene element 212, as shown in step 312, the
`2D-to-3D converter 110 can execute a blurring algorithm, as
`shoWn in step 314. The blurring algorithm can blur the offset
`map to correct any perceptual inaccuracies associated With
`the