Point reposition depth mapping
View Patent ↗A method and apparatus for providing optimal correction to depth mapping between captured and displayed stereoscopic content. The solution is derived in a continuous form that can be implemented through CGI scaling techniques compatible with image rendering techniques. Similar correction can be implemented with variable depth-dependent camera separation and disparity re-mapping. The latter is applicable to correcting existing stereoscopic content.
1. A method for processing a Computer Generated (CG) stereoscopic scene, the method comprising:
receiving a CG model of a stereoscopic scene having left image and right image model points; and
determining repositioning for the left image and right image model points in the CG model from input variables comprising an input disparity parameter pair and a scene depth parameter pair;
wherein the scene depth parameter pair comprises a depth of the furthest object in the CG stereoscopic scene and a depth of the closest object in the CG stereoscopic scene, further wherein the depths of the furthest object and the closest object in the CG stereoscopic scene are determined from the CG stereoscopic scene; and
wherein the depths of the closest object and the furthest object in the CG stereoscopic scene remain substantially unchanged after repositioning;
wherein the input disparity parameter pair comprises D max and D min , wherein D max is the maximum disparity limit for parallel eyes, and D min is the maximum disparity limit for crossed eyes,
wherein the scene depth parameter pair comprises Z max and Z min , wherein Z max is the furthest object in the CG stereoscopic scene, and Z min is the closest object in the CG stereoscopic scene;
wherein determining repositioning for left image and right image model points comprises using a point reposition depth mapping function comprising:
Z
′
(
Z
)
=
(
α
·
ln
(
Z
+
β
)
+
B
)
·
β
1
-
α
·
ln
(
Z
+
β
)
-
B
,
wherein
β
=
(
D
max
-
D
min
)
·
(
Z
min
·
Z
max
)
[
E
·
(
Z
max
-
Z
min
)
-
D
max
·
Z
max
+
D
min
·
Z
min
]
,
wherein
α
=
β
·
(
Z
min
-
Z
max
)
ln
[
(
Z
min
+
β
)
Z
max
+
β
)
]
·
[
(
Z
min
+
β
)
·
(
Z
max
+
β
)
]
,
wherein
B
=
Z
min
Z
min
+
β
-
α
·
ln
(
Z
min
+
β
)
,
wherein Z is an initial model point location from a left-eye or a right-eye observation point, wherein
Z
=
(
x
y
z
)
,
and
wherein E is eye separation.
2. The method of claim 1 , further comprising determining the input disparity parameter pair from the CG model.
3. The method of claim 1 , further comprising determining the input disparity parameter pair from a predetermined setting.
4. The method of claim 1 , further comprising determining the scene depth parameter pair from the CG model.
5. The method of claim 1 , further comprising determining the scene depth parameter pair from a predetermined setting.
6. The method of claim 1 , wherein determining repositioning for left image and right image model points comprises using a point reposition depth mapping function comprising:
Z
′
(
Z
)
=
β
(
K
·
Z
-
γ
-
1
)
,
wherein
β
=
(
D
max
-
D
min
)
·
(
Z
min
·
Z
max
)
[
E
·
(
Z
max
-
Z
min
)
-
D
max
·
Z
max
+
D
min
·
Z
min
]
,
wherein
γ
=
ln
[
(
E
-
D
max
)
(
E
-
D
min
)
]
ln
(
Z
min
Z
max
)
,
wherein
K
=
(
Z
min
+
β
)
Z
min
1
-
γ
,
wherein Z is an initial model point location from a left-eye or a right-eye observation point, wherein
Z
=
(
x
y
z
)
,
and
wherein E is eye separation.
7. The method of claim 1 , wherein the input disparity parameter pair is based on a minimum separation and a maximum separation.
8. The method of claim 1 , wherein determining repositioning comprises referencing a predetermined point reposition depth mapping from a look up table based on the input variables.
9. The method of claim 1 , wherein the input variables further comprise a head tilt variable, wherein the stereoscopic CG scene is modified in accordance with the head tilt variable.
10. The method of claim 1 , further comprising rendering the CG modified stereoscopic scene.
11. An apparatus for processing a computer generated (CG) stereoscopic scene comprising at least one processor implementing:
receiving a CG model of a stereoscopic scene having left image and right image model points; and
repositioning the left image and right image model points in the CG model from input variables comprising an input disparity parameter pair and a scene depth parameter pair,
wherein the scene depth parameter pair comprises the furthest object in the CG stereoscopic scene and the closest object in the CG stereoscopic scene, and
wherein the left and right image model points of the furthest object and the closest object in the CG stereoscopic scene remains substantially unchanged after repositioning further wherein the depths of the furthest object and the closest object in the CG stereoscopic scene are determined from the CG stereoscopic scene;
wherein the input disparity parameter pair comprises D max and D min , wherein D max is the maximum disparity limit for parallel eyes, and D min is the maximum disparity limit for crossed eyes,
wherein the scene depth parameter pair comprises Z max and Z min , wherein Z max is the furthest object in the CG stereoscopic scene, and Z min is the closest object in the CG stereoscopic scene,
wherein the processing subsystem utilizes a point reposition depth mapping function comprising:
Z
′
(
Z
)
=
(
α
·
ln
(
Z
+
β
)
+
B
)
·
β
1
-
α
·
ln
(
Z
+
β
)
-
B
,
wherein
β
=
(
D
max
-
D
min
)
·
(
Z
min
·
Z
max
)
[
E
·
(
Z
max
-
Z
min
)
-
D
max
·
Z
max
+
D
min
·
Z
min
]
,
wherein
α
=
β
·
(
Z
min
-
Z
max
)
ln
[
(
Z
min
+
β
)
Z
max
+
β
)
]
·
[
(
Z
min
+
β
)
·
(
Z
max
+
β
)
]
,
wherein
B
=
Z
min
Z
min
+
β
-
α
·
ln
(
Z
min
+
β
)
,
wherein Z is an initial model point location from a left-eye or a right-eye observation point, where
Z
=
(
x
y
z
)
,
and
wherein E is eye separation.
12. The apparatus of claim 11 further comprises a buffer.
13. The apparatus of claim 11 , wherein the processor utilizes a point reposition depth mapping function comprising:
Z
′
(
Z
)
=
β
(
K
·
Z
-
γ
-
1
)
,
wherein
β
=
(
D
max
-
D
min
)
·
(
Z
min
·
Z
max
)
[
E
·
(
Z
max
-
Z
min
)
-
D
max
·
Z
max
+
D
min
·
Z
min
]
,
wherein
γ
=
ln
[
(
E
-
D
max
)
(
E
-
D
min
)
]
ln
(
Z
min
Z
max
)
,
wherein
K
=
(
Z
min
+
β
)
Z
min
1
-
γ
,
wherein Z is an initial model point location from a left-eye or a right-eye observation point, wherein
Z
=
(
x
y
z
)
,
and
wherein E is eye separation.
14. The apparatus of claim 11 , wherein the input variables further comprise a head tilt variable, wherein the stereoscopic CG scene is modified in accordance with the head tilt variable.