WO1995004331A1 - Three-dimensional image synthesis using view interpolation - Google Patents

Three-dimensional image synthesis using view interpolation Download PDF

Info

Publication number
WO1995004331A1
WO1995004331A1 PCT/US1994/008467 US9408467W WO9504331A1 WO 1995004331 A1 WO1995004331 A1 WO 1995004331A1 US 9408467 W US9408467 W US 9408467W WO 9504331 A1 WO9504331 A1 WO 9504331A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
images
pixels
scene
pixel
Prior art date
Application number
PCT/US1994/008467
Other languages
French (fr)
Inventor
Shenchang Eric Chen
Lance Williams
Original Assignee
Apple Computer, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Apple Computer, Inc. filed Critical Apple Computer, Inc.
Priority to AU73743/94A priority Critical patent/AU7374394A/en
Publication of WO1995004331A1 publication Critical patent/WO1995004331A1/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T15/003D [Three Dimensional] image rendering
    • G06T15/10Geometric effects
    • G06T15/20Perspective computation
    • G06T15/205Image-based rendering
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2210/00Indexing scheme for image generation or computer graphics
    • G06T2210/44Morphing

Definitions

  • the present invention is directed to the computer generation of three-dimensional graphic scenes from different perspectives, and more particularly to the generation of such scenes through the synthesis of two-dimensional images.
  • the ability to generate a large number of images of an environment from closely spaced viewpoints is a very useful capability.
  • a traditional application of this capability is a flight in the cabin of an aircraft simulator.
  • a more contemporary application is a walk through a virtual environment. In both cases the same scene is displayed from the view of a virtual camera controlled by the user. A different view of a scene must be determined and presented for each different location of the camera.
  • a typical approach to this objective is to rely on a computer to repetitively render the scene from the different viewpoints. This approach has two major drawbacks.
  • a number of approaches have been proposed to address this problem. Most of these approaches use a preprocess technique to compute a subset of the scene visible from a specified viewing region. Only the potentially visible objects are processed in the available time. This approach does not completely solve the problem, however, because there may be viewing regions from which all objects are visible.
  • a different method has been developed to approximate the visibility at a location from adjacent environment maps.
  • the environment maps are Z- buffered images rendered from a set of discrete viewpoints in a three-dimensional space. Each environment map shows a complete view of the scene from a point.
  • An environment map can take the form of a cubic map, computed by rendering a cube of 90° views radiating from the point.
  • the environment maps are pre-computed and stored with viewpoints arranged in a structured way, such as a three-dimensional lattice.
  • An image from a new viewpoint can be generated by re-sampling the environment maps stored in adjacent locations.
  • the re-sampling process involves rendering the pixels in the environment maps as three-dimensional polygons from the new viewpoint.
  • the rendering time is proportional to the environment map resolutions and is independent of the scene complexity.
  • the requirement that samples from the maps be regenerated as three-dimensional polygons imposes substantial overhead, and may require special-purpose hardware.
  • these objectives are achieved through the use of view interpolation to generate intermediate images from stored images that are taken from nearby viewpoints.
  • the view interpolation is based on techniques that are similar to image morphing.
  • a plurality of images are stored which respectively correspond to the views of a scene from different locations.
  • adjacent images are morphed to create the new image.
  • Maps which identify the correspondence between any two adjacent images are pre-computed and stored. These maps are employed during the morphing process, to thereby allow the image generation to be carried out at interactive rates.
  • a block compression technique is employed to compress the maps, thereby reducing the number of computations that are performed in the generation of a new image.
  • the visibility priority of pixels within an image can be made view independent, thereby eliminating the need to resolve visibility during the morphing step.
  • mapping operation enables other results to be achieved more rapidly as well.
  • Many computer graphic renderings are computed much more slowly than real time, but rely on the capability of creating many subsidiary images from closely spaced viewpoints to depict complex effects of light and shadow in a single final image.
  • the pictures created by the mapping techniques of the present invention can be used instead of much more expensive traditional rendering.
  • a sequence of mapped views can be constructed to portray soft shadows that result in a scene from an extended light source.
  • blurring due to motion of an object within a scene can be computed more efficiently.
  • Figure 1 is a perspective view of an object constituting a scene to be viewed, along with a path for various viewpoints;
  • Figure 2 is a block diagram of a system for presenting views in accordance with the present invention.
  • Figure 3 is an illustration of three stored views of the scene of Figure 1, showing the correspondence maps that are generated for the stored views;
  • Figures 4A-4C depict the sequence of general steps that are carried out in the morphing of an image
  • Figures 5A-5C show a sequence of related images, illustrating holes that can result during image interpolation
  • Figure 6 is a ray trace diagram which depicts the manner in which holes can be formed in an image
  • Figure 7 is an illustration of a scene that is viewed from three respective viewpoints, depicting the feature of view- independent visibility priority
  • Figure 8 is an illustration of an image frame, depicting quadtree blocks;
  • Figures 9A and 9B are flowcharts of the main steps involved in the implementation of the present invention.
  • Figure 10 is a graph of a portion of an image array, depicting the locations of different viewpoints
  • Figure 11A and 11B are graphs of a temporal visibility function for a pixel in a target image.
  • Figure 12 is an illustration of the application of the invention to the generation of soft shadows.
  • the present invention is directed to the generation of images representing the views of a three- dimensional scene from different perspectives, such as different locations of a virtual camera.
  • the camera might be movable in one-dimensional, two- dimensional or three-dimensional space.
  • the scene that is being viewed is assumed to be static, and all changes which occur in the scene are a result of camera movement.
  • the scene to be viewed could comprise a single, three- dimensional object, such as a teapot 10 shown in Figure 1.
  • a ring 12 surrounding the object 10 illustrates the path of movement of a virtual camera through which the scene is viewed. Three positions along this path, displaced by 90° from one another, are respectively indicated by the arrows labelled A, B and C.
  • the viewed scene is actually a two-dimensional image that is presented on a suitable display 14, such as a monitor for a computer system, a television set, or any other suitable display arrangement.
  • the generation of the image is carried out under the control of a computer 16.
  • the movement of the virtual camera, to present different views of the scene is controlled by the viewer through a suitable input device 18.
  • This input device could comprise, for example, a joystick in a simulator, a head position detector in a virtual reality system, cursor control keys on a keyboard, or a mouse. If the viewer inputs a command for the camera to move to the right, the viewpoint will move from position A, for example, toward position B.
  • the image presented on the display 14 changes along with the movement of the camera, i.e. objects in the scene will move in the opposite direction. In the case of the teapot 10 shown in Figure 1, it will appear to rotate in a direction from right to left.
  • FIG. 1 selected views of a scene, from different viewpoints, are recorded.
  • the view of the object 10 from each of the positions A, B and C is stored in a memory of the computer 16.
  • FIG 3 Each of these three views is represented in Figure 3.
  • These views can be computer-generated images, artists' renditions, or digitized versions of respective photographic images.
  • the image from any viewpoint between any two of the recorded locations can be obtained by interpolation of two adjacent stored images.
  • the user's view of the scene is not limited to the prestored images. Rather, the scene can be viewed from any viewpoint within a continuous range defined by the stored images, providing a more realistic presentation.
  • the interpolation is carried out through image morphing.
  • image morphing is a simultaneous interpolation of shape and texture.
  • the morphing technique involves two basic steps. In the first step, a correspondence is established between two images. This step is usually the most difficult part of the morphing technique.
  • the correspondence is established by a human operator. The operator might, for example, define a set of corresponding points or line segments within a pair or set of images. An algorithm is then employed to determine the correspondence, or mapping, of the remaining points of the images.
  • the second basic step in the process is to employ the correspondence mapping to interpolate the shape of each image toward the other, according to the particular intermediate image to be formed.
  • FIG. 4A a pair of images are respectively labelled A and B.
  • a pixel 20, or group of pixels is identified in one of the images, and the location of the same pixel or group of pixels is determined in the second image.
  • pixel 20 in image A is located in image B at the position identified by the reference character 22.
  • the displacement of the pixel 20 in going from image A to image B is used to form an offset map M ⁇ .
  • the offset map is comprised of vectors, such as the vector 24, which identify the displacement of each pixel in going from the source image (A) to a destination image (B).
  • an interpolated offset map is determined for an intermediate viewing position that is located between the respective viewpoints for the images A and B.
  • the product of the offset map M, b and a coefficient t is computed to determine an interpolated offset map M at .
  • the coefficient t represents the location of the viewing position for which the interpolated map is computed.
  • the coefficient t can be the ratio of (a) the distance between the intermediate viewing position and the viewing position for a source image A, relative to (b) the total distance between the viewing positions for the source image A and a destination image B, respectively.
  • each of the images A, B and C from various viewpoints is stored in the memory of the computer 16.
  • the offset maps which describe the correspondence between each two adjacent images are also precomputed and stored.
  • a pair of bi ⁇ directional maps are stored for each pair of adjacent images. Referring again to Figure 3, one map 19a identifies the correspondence between images in going from image A to image B, and the other map 19b identifies the correspondence in going from image B to image A.
  • each of the two images A and B can act as a source image and a destination image.
  • two or more pixels in the source image may map to the same pixel location in the interpolated image.
  • the z- coordinate of each of these pixels can be compared to one another, and only the pixel having the nearest (most proximal) z-coordinate (as measured from the location of the camera) is displayed.
  • two deformed images are computed, i.e., one going from the source image to the destination image and one going from the destination image to the source image. Once these two deformed images have been determined, they are crossed-dissolved to produce the final interpolated image.
  • the morphing occurs between different views of the same scene or object, rather than between completely different images, and the surface shading of the same object is view independent. As a result, there is no strict need to cross-dissolve plural deformed images. However, cross-dissolving can still be employed where the z values are identical (or within a threshold distance), to provide smoother interpolation.
  • FIG. 5A-5C illustrate a scene that is viewed from different viewpoints respectively moving to the right.
  • the shaded regions 24 in Figures 5B and 5C indicate holes, i.e. pixels in the interpolated images that have no correspondence in the original source image.
  • One method to fill the holes is to interpolate the colors or offset vectors of adjacent pixels.
  • the destination image can be first filled with a predetermined background color. This background color is "reserved" as a marker, and is restricted from all prestored source images.
  • the hole pixels will retain the background color. New colors for these hole pixels can be computed by interpolating the colors of non-background pixels that are adjacent to or otherwise in the neighborhood of the hole pixels.
  • the offset vectors of the neighborhood pixels can be interpolated. The interpolated offset can be used with reference back to the source image, to obtain a new sample color.
  • Holes can also arise from objects that are invisible in each of the source images but visible in the destination image.
  • An example is shown in Figure 6.
  • two large objects 26 appear in the foreground of the scene, and a smaller object 28 is located behind them, in a hole region that is not visible from either of the source viewpoints A and B.
  • the hole region of concern is the intersection of the umbra regions Ua, Ub cast by each of the viewpoints A and B, and the visible region Vm from an intermediate viewpoint M.
  • the small object 28 in the hole region is completely missed by the two source images from points A and B, although it should be visible from the viewpoint M.
  • the problem of holes such as illustrated in Figure 6, can be reduced by using multiple source images to minimize the umbra regions. The closer the two source viewpoints are placed to one another, the greater the reduction in the number of holes.
  • FIG. 7 illustrates a scene containing two objects 30 and 32. This scene is viewed from three locations. In view 1, the viewing priority is that point A on object 30 is closer than point B on object 32. This priority is also valid for view 2, since points A and B do not overlap in this view. However, the priority is not correct for view 3, where points A and B overlap.
  • the field of view from each viewpoint is indicated by a triangle 34.
  • An angle ⁇ is formed between the edge of one viewpoint's field of view and the center line (or optical axis) for the adjacent viewpoint. It has been determined that, as long as the angle ⁇ is less than 90 degrees for adjacent viewpoints, the A-B priority of any two pixels does not need to be changed when moving from one viewpoint to the other. Thus, a view-independent visibility priority can be established for every source pixel for a given viewing range, by selecting the source images so that the angle ⁇ is less than 90° between any two adjacent images.
  • the pixels are ordered from back to front based on their original z coordinates. Subsequently, during the interpolation process, the pixels are displayed in back-to-front order. This ordering of the pixels eliminates the need to interpolate the z coordinate of each pixel during the interpolation process. Since adjacent pixels tend to move together in the mapping from one image to another, a block compression technique, such as quadtree, can be employed to compress the offset maps 19. Adjacent pixels which move in a similar manner can be grouped in blocks 36, as shown in Figure 8, and moved together. This compression serves two purposes. First, it reduces the size of the offset maps. Second, it allows offsets for entire blocks of pixels to be interpolated, rather than pixel-by-pixel. This second aspect of the block compression greatly accelerates the interpolation process, since the main effort in the process is the interpolation of the offset vectors.
  • the compression ratio that is employed can be related to the complexity of the image depth and the viewpoint movement. For images with high depth complexity, the compression ratio is preferably low. The ratio should also be lower as the movement of the viewpoint results in greater pixel depth change.
  • a maximum offset threshold can be employed to determine the appropriate compression ratio.
  • a maximum offset threshold of one or two pixels, for example, would mean that the offset vector coordinates for individual pixels within a block do not differ by more than one or two pixel units.
  • the various blocks can be of differing sizes, in dependence upon the variation in the magnitudes of the offset vectors.
  • the threshold provides a smooth quality degradation path for increased performance. Large threshold factors result in fewer quadtree blocks and, therefore, reduce the interpolation time.
  • the view-independent visibility priority described above is applicable to the quadtree blocks.
  • the priority can be assigned to every quadtree pixel block. Since the pixels within a block have similar offsets, they also have similar z coordinates. The z coordinates within a block can be averaged, or otherwise filtered, to determine a z-value for the block as a whole. This value is employed in a visibility priority sort. A sorted list of pixel blocks can thereby be created, which is the value for the entire range between two stored views.
  • the process of the present invention involves two main stages, a preprocessing stage and an interactive interpolation stage.
  • the preprocessing stage the correspondence between each adjacent pair of source and destination images is established.
  • This stage is the more computationally intensive, and can be carried out in advance, using as much time as is necessary.
  • the interpolation stage a new image is generated to correspond to a new viewing location. This stage is implemented in real time, at interactive rates.
  • the steps of the preprocessing stage are depicted in Figure 9A.
  • a first step 40 the appropriate input data is gathered and/or determined.
  • This input data includes the source and destination images, as well as range data (z coordinates) and camera parameters associated with each image.
  • the input data also includes a threshold factor for the quadtree decomposition.
  • an offset map is created from the source to the destination image.
  • the offset map is decomposed into quadtree blocks, and the blocks are added to a block list (step 44).
  • the block list is then sorted from back to front, according to each block's respective z coordinates (step 46).
  • steps are repeated for each set of adjacent stored images. For every pair of images, the steps are carried out twice, once in each direction.
  • step 50 input data, comprising interpolation parameters and a sorted block list, are obtained.
  • the interpolation parameters are the parametric coordinates of the new viewing location, relative to the viewing locations of adjacent stored images.
  • the interpolated image is filled with a distinctive background color.
  • a new location is computed for each block in the block list, going from back to front (step 54). As the location of each block is determined, it is copied to that location in the interpolated image (step 56). After all of the blocks have been copied to the new image, each pixel which retains the background color is identified (step 57), and its color is computed by filtering the colors of adjacent non-background pixels, or the offset vectors of the adjacent pixels (step 58).
  • the movement of the virtual camera is one-dimensional.
  • the user's control over the displayed scene can be multi ⁇ dimensional.
  • the views of the teapot 10 can be taken from any point on a sphere which completely surrounds the teapot, rather than only a ring.
  • An array of prestored images is determined for selected points on the sphere. These images can be connected to form a graph, where each node on the graph comprises a source image and each arc in the graph (connecting any two adjacent nodes) represents the correspondence mapping between the images. Since the mapping is bidirectional, two maps are associated with each arc.
  • FIG. 10 An example of a portion of an image array is shown in Figure 10.
  • the images are grouped in sets of three, so that the arcs 60 between adjacent image nodes 62 form triangles.
  • a scene can be viewed from any point encompassed by the triangle.
  • the interpolation is carried out with two offset maps from two source images, as explained with reference to the example of Figure 1.
  • a viewpoint 66 located within the triangle its barycentric coordinates are determined, and used to interpolate from among three source images.
  • each pixel in image A will have two offset vectors associated with it.
  • One vector identifies that pixel's displacement in going from image A to image B, and the other vector corresponds to the pixel's displacement in going from image A to image C.
  • the barycentric coordinates of the viewpoint 66 determine the coefficients by which each of these vectors is scaled.
  • Each scaled vector represents an interpolated image, and all of the interpolated images (in this case, three) are combined to form a composite image that is displayed.
  • the changes in the displayed scene can be multi-dimensional and are determined by multivariate interpolation.
  • the dimensions can be other than spatial, e.g. they can relate to object movement or surface distortions.
  • the changes in each dimension are accompanied by an appropriate correspondence map, which is employed in the multivariate interpolation.
  • the view-independent visibility priority may no longer app y * In such a case, it may be appropriate to include range data with each sample or block of samples, to effect compositing of images.
  • the ability of the present invention to rapidly generate a large number of scenes facilitates the simulation of motion in a displayed scene. For example, if the camera's location is moved at a rapid rate some of the objects in the scene may appear blurred. In the present invention, such blurring can be simulated through a temporal sampling technique. If an image of a motion sequence is rendered for an instant of time, rather than a time interval, the motion will appear to be jerky, and the image is said to be aliased in the temporal domain.
  • One conventional technique to perform temporal anti-aliasing is super-sampling. The motion is sampled at a higher rate in the temporal domain, and the samples are filtered to the display rate. Super-sampling requires the computation of many more samples. For images which require a significant amount of processing to render, this technique can be computationally expensive.
  • the morphing method of the present invention allows additional temporal samples to be created by interpolation.
  • the interpolation time is constant regardless of the complexity of the images being interpolated.
  • the sampling rate is determined by the largest offset vector from the morph map, in order to perform proper anti-aliasing.
  • the sampling rate is determined on the basis of the worst case. If an image has a fast moving object and slowly moving background, the super- sampling approach is not very efficient.
  • One way to improve upon this approach is to segment the images based on object movement, and use different sampling rates for each segment. For instance, a foreground object in a scene should be sampled at a high rate, while a stationary wall in the background needs only a few samples.
  • the different segments can be sorted in order of depth. Each segment is filtered independently and a temporal coverage value is stored for each pixel, to indicate the ratio of background samples relative to all samples.
  • the multiple segment layers are composited in front- to-back order, with each segment's pixel colors attenuated by the coverage value.
  • the function of the coverage value therefore, is to normalize the brightness of all of the pixels, so that the pixels which have been sampled a greater number of times do not have their brightnesses artificially increased.
  • Another approach to anti-aliasing motion is to generate a continuous temporal visibility function for each pixel from the mapping.
  • this continuous function describes when and which source pixels cross the window over a predetermined time interval.
  • the continuous function can be computed by moving all of the source image pixels to their destinations, and recording the time they cross each target pixel's center. After this process, each target pixel has a list of records for the source pixels which have crossed the target pixel in the time interval. As shown in Figure 11 A, each record contains an identification of the source pixel's coordinates, Pxy, and a sub-interval for the crossing times. This list is then sorted by the z-coordinates, and occluded sub- intervals are deleted, as shown in Figure 11B. The remaining list is a continuous function and can be integrated for any time interval to simulate motion.
  • Another application of the present invention is in the rendering of shadows.
  • a shadow buffer algorithm was computed using a z-buffer, or shadow map, from the point of view of a light source.
  • To compute shadows the coordinates of a surface point in an image were transformed to the light source's space, and its z-coordinates compared to the corresponding z-coordinates in the shadow map. While this approach is acceptable for point light sources, it is not very efficient for the approximation of linear or area sources, which require many different point light sources to be employed.
  • the morphing method of the present invention can be used to significantly reduce the cost of computing a shadow map for each of the point sources.
  • a shadow map is computed first for each of the two end points lightl and light2 of a linear light source, using a conventional rendering method.
  • Interpolation maps, mapl and map2, from the viewpoint to each of the two end points are also computed, to transform the scene coordinates to each point source's coordinate space.
  • a shadow map for an intermediate point, light3, on the linear source is interpolated from the end point shadow maps, using the morphing method.
  • the same interpolation factor is used to inte ⁇ olate the two interpolation maps, to create a map from the viewpoint to the intermediate light source point.
  • a standard shadow buffer algorithm can then be employed to compute shadows for the intermediate point source. This process is repeated for all intermediate points at a desired interval. The resulting shadow images are composited, to create a soft shadow of the linear source.
  • the present invention employs images and environment maps to represent a three- dimensional scene. To view the scene from different perspectives, images adjacent to a viewpoint are interpolated to create the desired view. With this method, only a few key images are required, and interpolation is used to generate in-between frames at interactive rates, regardless of scene complexity. This ability to generate images at an accelerated rate is a result of separating the slower process of computing image correspondence from the faster interpolation process, and carrying out the slow processes ahead of time.
  • the image mo ⁇ hing method of the present invention is computationally inexpensive and yet provides the capability to present images from any viewpoint in a defined range at a real-time rate. While a large amount of data may need to be stored, the amount needed to compute any given frame is relatively small and can be read from secondary storage, as needed. Thus, the approach of the present invention is quite appropriate for CD-ROM based devices, for example, due to their large storage capacity. As the complexity of geometric models increases, the advantages provided by the image-based approach of the present invention become more significant.
  • maps may store new image coordinates for pixels in a source image, or offset vectors to these new image coordinates (from each pixel's original coordinates). Such a set of coordinates or offsets is termed a "forward mapping", as it maps each pixel to a new target location.
  • a forward mapping Such a set of coordinates or offsets is termed a "forward mapping", as it maps each pixel to a new target location.
  • An inverse map is a set of coordinates or offsets, defined for each pixel in the target image, which specifies the location of pixels in a source image.
  • the forward map from A to B indicates where to move each pixel in A to arrive at a close approximation to image B.
  • a new location is specified for every source pixel in such a mapping, but it does not guarantee that all target image pixels will be covered with a pixel from the source image, and so (as previously discussed) "holes" may arise.
  • the inverse map from A to B indicates, for each pixel in image B, the coordinates (or offsets) of corresponding pixels in image A. In using the inverse map, no holes occur, since every target pixel has a defined source pixel in the undeformed image.
  • the regions of the maps outside the boundaries of objects should contain coordinates or offsets extrapolated from the coordinates or offsets within object boundaries.
  • the boundaries themselves, which delimit the visible portions of objects can be represented separately, as mattes.
  • a matte is an extra component of the image - along with the red, green, and blue components - which defines the opacity of the image at each point.
  • the mattes for the source and target images in a mo ⁇ h are indexed, deformed, and composited in the same way as the red, green, and blue components of the images.

Abstract

Three-dimensional scenes are portrayed from different viewpoints by morphing two-dimensional images. Various key views of a scene are stored, along with offset maps that identify the correspondence of pixels in adjacent stored images. When an intermediate view of the scene is to be presented, one ore more stored views are interpolated through a morphing technique. Since the key views and their offset data are pre-stored, the morphing and presentation of new views can be carried out at interactive rates. By providing the ability to quickly compute many closely spaced views, the disclosed morphing technique also facilitates the rapid computation of soft shadows and motion blur in images.

Description

THREE-DIMENSIONAL IMAGE SYNTHESIS USING VIEW INTERPOLATION
Field of the Invention
The present invention is directed to the computer generation of three-dimensional graphic scenes from different perspectives, and more particularly to the generation of such scenes through the synthesis of two-dimensional images.
Background Information
The ability to generate a large number of images of an environment from closely spaced viewpoints is a very useful capability. A traditional application of this capability is a flight in the cabin of an aircraft simulator. A more contemporary application is a walk through a virtual environment. In both cases the same scene is displayed from the view of a virtual camera controlled by the user. A different view of a scene must be determined and presented for each different location of the camera. The computation of global illumination effects pertinent to the scene, such as shadows, diffuse and specular inter-reflections, also requires a large number of visibility calculations. A typical approach to this objective is to rely on a computer to repetitively render the scene from the different viewpoints. This approach has two major drawbacks. First, real-time rendering of a complex scene is computationally expensive and usually requires specialized graphics hardware. Second, the rendering time is usually not constant and is dependent on the scene complexity. This problem is particularly critical in simulation and virtual reality applications because of the demand for real-time feedback. Since scene complexity is potentially unbounded, this latter problem will always exist regardless of the available processing power.
A number of approaches have been proposed to address this problem. Most of these approaches use a preprocess technique to compute a subset of the scene visible from a specified viewing region. Only the potentially visible objects are processed in the available time. This approach does not completely solve the problem, however, because there may be viewing regions from which all objects are visible. A different method has been developed to approximate the visibility at a location from adjacent environment maps. In this method, the environment maps are Z- buffered images rendered from a set of discrete viewpoints in a three-dimensional space. Each environment map shows a complete view of the scene from a point. An environment map can take the form of a cubic map, computed by rendering a cube of 90° views radiating from the point. The environment maps are pre-computed and stored with viewpoints arranged in a structured way, such as a three-dimensional lattice. An image from a new viewpoint can be generated by re-sampling the environment maps stored in adjacent locations. The re-sampling process involves rendering the pixels in the environment maps as three-dimensional polygons from the new viewpoint. In this approach, the rendering time is proportional to the environment map resolutions and is independent of the scene complexity. However, the requirement that samples from the maps be regenerated as three-dimensional polygons imposes substantial overhead, and may require special-purpose hardware.
Accordingly, it is desirable to provide a technique for rapidly generating images from nearby viewpoints which is independent of scene complexity and which does not require specialized hardware to implement, to enable a continuous sequence of new images to be generated rapidly on a real-time basis.
Summary of the Invention
In accordance with the present invention, these objectives are achieved through the use of view interpolation to generate intermediate images from stored images that are taken from nearby viewpoints. The view interpolation is based on techniques that are similar to image morphing.
More particularly, a plurality of images are stored which respectively correspond to the views of a scene from different locations. To generate the view from an intermediate viewpoint between those locations, adjacent images are morphed to create the new image. Maps which identify the correspondence between any two adjacent images are pre-computed and stored. These maps are employed during the morphing process, to thereby allow the image generation to be carried out at interactive rates. To accelerate the morphing process, a block compression technique is employed to compress the maps, thereby reducing the number of computations that are performed in the generation of a new image. In addition, though proper selection of the stored images, the visibility priority of pixels within an image can be made view independent, thereby eliminating the need to resolve visibility during the morphing step.
The mapping operation enables other results to be achieved more rapidly as well. Many computer graphic renderings are computed much more slowly than real time, but rely on the capability of creating many subsidiary images from closely spaced viewpoints to depict complex effects of light and shadow in a single final image. In such situations, the pictures created by the mapping techniques of the present invention can be used instead of much more expensive traditional rendering. For example, a sequence of mapped views can be constructed to portray soft shadows that result in a scene from an extended light source. In addition, blurring due to motion of an object within a scene can be computed more efficiently.
These, as well as other features and advantages of the invention are explained in greater detail hereinafter with reference to specific examples illustrated in the accompanying drawings. Brief Descriptions of the Drawings
Figure 1 is a perspective view of an object constituting a scene to be viewed, along with a path for various viewpoints;
Figure 2 is a block diagram of a system for presenting views in accordance with the present invention;
Figure 3 is an illustration of three stored views of the scene of Figure 1, showing the correspondence maps that are generated for the stored views;
Figures 4A-4C depict the sequence of general steps that are carried out in the morphing of an image;
Figures 5A-5C show a sequence of related images, illustrating holes that can result during image interpolation;
Figure 6 is a ray trace diagram which depicts the manner in which holes can be formed in an image;
Figure 7 is an illustration of a scene that is viewed from three respective viewpoints, depicting the feature of view- independent visibility priority;
Figure 8 is an illustration of an image frame, depicting quadtree blocks; Figures 9A and 9B are flowcharts of the main steps involved in the implementation of the present invention;
Figure 10 is a graph of a portion of an image array, depicting the locations of different viewpoints;
Figure 11A and 11B are graphs of a temporal visibility function for a pixel in a target image; and
Figure 12 is an illustration of the application of the invention to the generation of soft shadows.
Detailed Description
As described previously, the present invention is directed to the generation of images representing the views of a three- dimensional scene from different perspectives, such as different locations of a virtual camera. Depending upon the particular application, the camera might be movable in one-dimensional, two- dimensional or three-dimensional space. To facilitate an understanding of the fundamental principles which underlie the present invention, it is first described with reference to examples of movement of the camera in one-dimensional and two-dimensional space. Furthermore, in these preliminary examples, the scene that is being viewed is assumed to be static, and all changes which occur in the scene are a result of camera movement. The scene to be viewed could comprise a single, three- dimensional object, such as a teapot 10 shown in Figure 1. A ring 12 surrounding the object 10 illustrates the path of movement of a virtual camera through which the scene is viewed. Three positions along this path, displaced by 90° from one another, are respectively indicated by the arrows labelled A, B and C.
In practice, with reference to Figure 2, the viewed scene is actually a two-dimensional image that is presented on a suitable display 14, such as a monitor for a computer system, a television set, or any other suitable display arrangement. The generation of the image is carried out under the control of a computer 16. The movement of the virtual camera, to present different views of the scene, is controlled by the viewer through a suitable input device 18. This input device could comprise, for example, a joystick in a simulator, a head position detector in a virtual reality system, cursor control keys on a keyboard, or a mouse. If the viewer inputs a command for the camera to move to the right, the viewpoint will move from position A, for example, toward position B. The image presented on the display 14 changes along with the movement of the camera, i.e. objects in the scene will move in the opposite direction. In the case of the teapot 10 shown in Figure 1, it will appear to rotate in a direction from right to left.
In accordance with the present invention, selected views of a scene, from different viewpoints, are recorded. In the example of Figure 1, the view of the object 10 from each of the positions A, B and C is stored in a memory of the computer 16. Each of these three views is represented in Figure 3. These views can be computer-generated images, artists' renditions, or digitized versions of respective photographic images. Once each of these views has been stored, the image from any viewpoint between any two of the recorded locations can be obtained by interpolation of two adjacent stored images. Thus, the user's view of the scene is not limited to the prestored images. Rather, the scene can be viewed from any viewpoint within a continuous range defined by the stored images, providing a more realistic presentation.
In accordance with the present invention, the interpolation is carried out through image morphing. Generally speaking, image morphing is a simultaneous interpolation of shape and texture. The morphing technique involves two basic steps. In the first step, a correspondence is established between two images. This step is usually the most difficult part of the morphing technique. Typically, the correspondence is established by a human operator. The operator might, for example, define a set of corresponding points or line segments within a pair or set of images. An algorithm is then employed to determine the correspondence, or mapping, of the remaining points of the images. The second basic step in the process is to employ the correspondence mapping to interpolate the shape of each image toward the other, according to the particular intermediate image to be formed. The morph is completed by blending the pixel values of the two images by the same respective coefficients. This technique, as applied to the present invention, is illustrated in Figures 4A-4C. In Figure 4A, a pair of images are respectively labelled A and B. In the first step of the image morphing process, a pixel 20, or group of pixels, is identified in one of the images, and the location of the same pixel or group of pixels is determined in the second image. In the example of Figure 4A, pixel 20 in image A is located in image B at the position identified by the reference character 22. The displacement of the pixel 20 in going from image A to image B is used to form an offset map M^. In essence, the offset map is comprised of vectors, such as the vector 24, which identify the displacement of each pixel in going from the source image (A) to a destination image (B).
In the second step of the process, an interpolated offset map is determined for an intermediate viewing position that is located between the respective viewpoints for the images A and B.
Referring to Figure 4B, the product of the offset map M,b and a coefficient t is computed to determine an interpolated offset map Mat. The coefficient t represents the location of the viewing position for which the interpolated map is computed. For example, the coefficient t can be the ratio of (a) the distance between the intermediate viewing position and the viewing position for a source image A, relative to (b) the total distance between the viewing positions for the source image A and a destination image B, respectively. Once the interpolated map MIt is determined, its vectors are applied to the pixels of the source image A, to result in an intermediate view T, as shown in Figure 4C. In other words, the pixels from the source image A are moved to their interpolated locations, as defined by the map M,t, to create the interpolated image T.
As noted previously, each of the images A, B and C from various viewpoints is stored in the memory of the computer 16. In addition, to reduce real-time processing requirements, the offset maps which describe the correspondence between each two adjacent images are also precomputed and stored. In practice, a pair of bi¬ directional maps are stored for each pair of adjacent images. Referring again to Figure 3, one map 19a identifies the correspondence between images in going from image A to image B, and the other map 19b identifies the correspondence in going from image B to image A. Thus, each of the two images A and B can act as a source image and a destination image.
In going from a source image to an interpolated image, two or more pixels in the source image may map to the same pixel location in the interpolated image. To resolve visibility, the z- coordinate of each of these pixels can be compared to one another, and only the pixel having the nearest (most proximal) z-coordinate (as measured from the location of the camera) is displayed.
In a conventional morphing process, two deformed images are computed, i.e., one going from the source image to the destination image and one going from the destination image to the source image. Once these two deformed images have been determined, they are crossed-dissolved to produce the final interpolated image. In the practice of the present invention, the morphing occurs between different views of the same scene or object, rather than between completely different images, and the surface shading of the same object is view independent. As a result, there is no strict need to cross-dissolve plural deformed images. However, cross-dissolving can still be employed where the z values are identical (or within a threshold distance), to provide smoother interpolation.
When one-directional mapping alone is used to compute the interpolated image, holes may occur in the final image. These holes are created at every target pixel to which no source sample is transformed. These holes appear first at the steep slopes in the depth map of an image as it is transformed, where adjacent pixels have widely differing depths, and where the mapping causes local image expansion. For example, Figures 5A-5C illustrate a scene that is viewed from different viewpoints respectively moving to the right. The shaded regions 24 in Figures 5B and 5C indicate holes, i.e. pixels in the interpolated images that have no correspondence in the original source image.
One method to fill the holes is to interpolate the colors or offset vectors of adjacent pixels. To identify holes, the destination image can be first filled with a predetermined background color. This background color is "reserved" as a marker, and is restricted from all prestored source images. After the inteφolated offset map is applied to the original image, the hole pixels will retain the background color. New colors for these hole pixels can be computed by interpolating the colors of non-background pixels that are adjacent to or otherwise in the neighborhood of the hole pixels. Alternatively, the offset vectors of the neighborhood pixels can be interpolated. The interpolated offset can be used with reference back to the source image, to obtain a new sample color.
Holes can also arise from objects that are invisible in each of the source images but visible in the destination image. An example is shown in Figure 6. In this example, two large objects 26 appear in the foreground of the scene, and a smaller object 28 is located behind them, in a hole region that is not visible from either of the source viewpoints A and B. The hole region of concern is the intersection of the umbra regions Ua, Ub cast by each of the viewpoints A and B, and the visible region Vm from an intermediate viewpoint M. The small object 28 in the hole region is completely missed by the two source images from points A and B, although it should be visible from the viewpoint M. The problem of holes, such as illustrated in Figure 6, can be reduced by using multiple source images to minimize the umbra regions. The closer the two source viewpoints are placed to one another, the greater the reduction in the number of holes.
As a further feature of the invention, certain criteria can be employed in the selection of views to eliminate, or at least substantially reduce, the need to interpolate and compare the z coordinates of each pixel to determine view priority. Figure 7 illustrates a scene containing two objects 30 and 32. This scene is viewed from three locations. In view 1, the viewing priority is that point A on object 30 is closer than point B on object 32. This priority is also valid for view 2, since points A and B do not overlap in this view. However, the priority is not correct for view 3, where points A and B overlap.
In the illustration of Figure 7, the field of view from each viewpoint is indicated by a triangle 34. An angle θ is formed between the edge of one viewpoint's field of view and the center line (or optical axis) for the adjacent viewpoint. It has been determined that, as long as the angle θ is less than 90 degrees for adjacent viewpoints, the A-B priority of any two pixels does not need to be changed when moving from one viewpoint to the other. Thus, a view-independent visibility priority can be established for every source pixel for a given viewing range, by selecting the source images so that the angle θ is less than 90° between any two adjacent images.
In the implementation of view-independent visibility priority, when the offset maps are created, the pixels are ordered from back to front based on their original z coordinates. Subsequently, during the interpolation process, the pixels are displayed in back-to-front order. This ordering of the pixels eliminates the need to interpolate the z coordinate of each pixel during the interpolation process. Since adjacent pixels tend to move together in the mapping from one image to another, a block compression technique, such as quadtree, can be employed to compress the offset maps 19. Adjacent pixels which move in a similar manner can be grouped in blocks 36, as shown in Figure 8, and moved together. This compression serves two purposes. First, it reduces the size of the offset maps. Second, it allows offsets for entire blocks of pixels to be interpolated, rather than pixel-by-pixel. This second aspect of the block compression greatly accelerates the interpolation process, since the main effort in the process is the interpolation of the offset vectors.
The compression ratio that is employed can be related to the complexity of the image depth and the viewpoint movement. For images with high depth complexity, the compression ratio is preferably low. The ratio should also be lower as the movement of the viewpoint results in greater pixel depth change. A maximum offset threshold can be employed to determine the appropriate compression ratio. A maximum offset threshold of one or two pixels, for example, would mean that the offset vector coordinates for individual pixels within a block do not differ by more than one or two pixel units. Thus, as shown in Figure 8, the various blocks can be of differing sizes, in dependence upon the variation in the magnitudes of the offset vectors. The threshold provides a smooth quality degradation path for increased performance. Large threshold factors result in fewer quadtree blocks and, therefore, reduce the interpolation time. The view-independent visibility priority described above is applicable to the quadtree blocks. The priority can be assigned to every quadtree pixel block. Since the pixels within a block have similar offsets, they also have similar z coordinates. The z coordinates within a block can be averaged, or otherwise filtered, to determine a z-value for the block as a whole. This value is employed in a visibility priority sort. A sorted list of pixel blocks can thereby be created, which is the value for the entire range between two stored views.
The foregoing aspects of the present invention are summarized in the flowcharts of Figures 9 A and 9B. Generally speaking, the process of the present invention involves two main stages, a preprocessing stage and an interactive interpolation stage. In the preprocessing stage, the correspondence between each adjacent pair of source and destination images is established. This stage is the more computationally intensive, and can be carried out in advance, using as much time as is necessary. In the interpolation stage, a new image is generated to correspond to a new viewing location. This stage is implemented in real time, at interactive rates.
The steps of the preprocessing stage are depicted in Figure 9A. As a first step 40, the appropriate input data is gathered and/or determined. This input data includes the source and destination images, as well as range data (z coordinates) and camera parameters associated with each image. The input data also includes a threshold factor for the quadtree decomposition. In the second step 42 of the process, an offset map is created from the source to the destination image. The offset map is decomposed into quadtree blocks, and the blocks are added to a block list (step 44). The block list is then sorted from back to front, according to each block's respective z coordinates (step 46).
These steps are repeated for each set of adjacent stored images. For every pair of images, the steps are carried out twice, once in each direction.
The interactive interpolation stage is illustrated in Figure
9B. As a first step 50, input data, comprising interpolation parameters and a sorted block list, are obtained. The interpolation parameters are the parametric coordinates of the new viewing location, relative to the viewing locations of adjacent stored images. In the second step 52, the interpolated image is filled with a distinctive background color. A new location is computed for each block in the block list, going from back to front (step 54). As the location of each block is determined, it is copied to that location in the interpolated image (step 56). After all of the blocks have been copied to the new image, each pixel which retains the background color is identified (step 57), and its color is computed by filtering the colors of adjacent non-background pixels, or the offset vectors of the adjacent pixels (step 58). In the example of Figure 1, the movement of the virtual camera is one-dimensional. In practical applications of the invention, the user's control over the displayed scene can be multi¬ dimensional. For example, the views of the teapot 10 can be taken from any point on a sphere which completely surrounds the teapot, rather than only a ring. An array of prestored images is determined for selected points on the sphere. These images can be connected to form a graph, where each node on the graph comprises a source image and each arc in the graph (connecting any two adjacent nodes) represents the correspondence mapping between the images. Since the mapping is bidirectional, two maps are associated with each arc.
An example of a portion of an image array is shown in Figure 10. In the graph, the images are grouped in sets of three, so that the arcs 60 between adjacent image nodes 62 form triangles. A scene can be viewed from any point encompassed by the triangle. For a viewpoint 64 located along one of the edges 60 of the triangle, the interpolation is carried out with two offset maps from two source images, as explained with reference to the example of Figure 1. For a viewpoint 66 located within the triangle, its barycentric coordinates are determined, and used to interpolate from among three source images.
More particularly, in the example of Figure 10, each pixel in image A will have two offset vectors associated with it. One vector identifies that pixel's displacement in going from image A to image B, and the other vector corresponds to the pixel's displacement in going from image A to image C. The barycentric coordinates of the viewpoint 66 determine the coefficients by which each of these vectors is scaled. Each scaled vector represents an interpolated image, and all of the interpolated images (in this case, three) are combined to form a composite image that is displayed.
From the foregoing, it can be seen that the changes in the displayed scene can be multi-dimensional and are determined by multivariate interpolation. Furthermore, the dimensions can be other than spatial, e.g. they can relate to object movement or surface distortions. The changes in each dimension are accompanied by an appropriate correspondence map, which is employed in the multivariate interpolation. As more dimensions are added, the view-independent visibility priority may no longer app y* In such a case, it may be appropriate to include range data with each sample or block of samples, to effect compositing of images.
The ability of the present invention to rapidly generate a large number of scenes facilitates the simulation of motion in a displayed scene. For example, if the camera's location is moved at a rapid rate some of the objects in the scene may appear blurred. In the present invention, such blurring can be simulated through a temporal sampling technique. If an image of a motion sequence is rendered for an instant of time, rather than a time interval, the motion will appear to be jerky, and the image is said to be aliased in the temporal domain. One conventional technique to perform temporal anti-aliasing is super-sampling. The motion is sampled at a higher rate in the temporal domain, and the samples are filtered to the display rate. Super-sampling requires the computation of many more samples. For images which require a significant amount of processing to render, this technique can be computationally expensive.
The morphing method of the present invention allows additional temporal samples to be created by interpolation. The interpolation time is constant regardless of the complexity of the images being interpolated. The sampling rate is determined by the largest offset vector from the morph map, in order to perform proper anti-aliasing.
When a super-sampling approach is employed, the sampling rate is determined on the basis of the worst case. If an image has a fast moving object and slowly moving background, the super- sampling approach is not very efficient. One way to improve upon this approach is to segment the images based on object movement, and use different sampling rates for each segment. For instance, a foreground object in a scene should be sampled at a high rate, while a stationary wall in the background needs only a few samples. In the case of motion caused by viewpoint changes, the different segments can be sorted in order of depth. Each segment is filtered independently and a temporal coverage value is stored for each pixel, to indicate the ratio of background samples relative to all samples. The multiple segment layers are composited in front- to-back order, with each segment's pixel colors attenuated by the coverage value. The function of the coverage value, therefore, is to normalize the brightness of all of the pixels, so that the pixels which have been sampled a greater number of times do not have their brightnesses artificially increased.
Another approach to anti-aliasing motion is to generate a continuous temporal visibility function for each pixel from the mapping. By treating a pixel in the destination image as a window, this continuous function describes when and which source pixels cross the window over a predetermined time interval. Referring to Figures 11A and 11B, the continuous function can be computed by moving all of the source image pixels to their destinations, and recording the time they cross each target pixel's center. After this process, each target pixel has a list of records for the source pixels which have crossed the target pixel in the time interval. As shown in Figure 11 A, each record contains an identification of the source pixel's coordinates, Pxy, and a sub-interval for the crossing times. This list is then sorted by the z-coordinates, and occluded sub- intervals are deleted, as shown in Figure 11B. The remaining list is a continuous function and can be integrated for any time interval to simulate motion.
Another application of the present invention is in the rendering of shadows. In a conventional approach, a shadow buffer algorithm was computed using a z-buffer, or shadow map, from the point of view of a light source. To compute shadows, the coordinates of a surface point in an image were transformed to the light source's space, and its z-coordinates compared to the corresponding z-coordinates in the shadow map. While this approach is acceptable for point light sources, it is not very efficient for the approximation of linear or area sources, which require many different point light sources to be employed.
The morphing method of the present invention can be used to significantly reduce the cost of computing a shadow map for each of the point sources. Referring to Figure 12, a shadow map is computed first for each of the two end points lightl and light2 of a linear light source, using a conventional rendering method. Interpolation maps, mapl and map2, from the viewpoint to each of the two end points are also computed, to transform the scene coordinates to each point source's coordinate space. A shadow map for an intermediate point, light3, on the linear source is interpolated from the end point shadow maps, using the morphing method. The same interpolation factor is used to inteφolate the two interpolation maps, to create a map from the viewpoint to the intermediate light source point. A standard shadow buffer algorithm can then be employed to compute shadows for the intermediate point source. This process is repeated for all intermediate points at a desired interval. The resulting shadow images are composited, to create a soft shadow of the linear source. From the foregoing, it can be seen that the present invention employs images and environment maps to represent a three- dimensional scene. To view the scene from different perspectives, images adjacent to a viewpoint are interpolated to create the desired view. With this method, only a few key images are required, and interpolation is used to generate in-between frames at interactive rates, regardless of scene complexity. This ability to generate images at an accelerated rate is a result of separating the slower process of computing image correspondence from the faster interpolation process, and carrying out the slow processes ahead of time.
The image moφhing method of the present invention is computationally inexpensive and yet provides the capability to present images from any viewpoint in a defined range at a real-time rate. While a large amount of data may need to be stored, the amount needed to compute any given frame is relatively small and can be read from secondary storage, as needed. Thus, the approach of the present invention is quite appropriate for CD-ROM based devices, for example, due to their large storage capacity. As the complexity of geometric models increases, the advantages provided by the image-based approach of the present invention become more significant.
One important design decision in the image moφhing method is the nature of the offset maps used to apply image deformations. In the foregoing discussion, maps may store new image coordinates for pixels in a source image, or offset vectors to these new image coordinates (from each pixel's original coordinates). Such a set of coordinates or offsets is termed a "forward mapping", as it maps each pixel to a new target location. Another possibility, often used in conventional image moφhing, is to use an "inverse mapping". An inverse map is a set of coordinates or offsets, defined for each pixel in the target image, which specifies the location of pixels in a source image. To illustrate, for any two images, A and B, for which correspondences can be mapped, the forward map from A to B indicates where to move each pixel in A to arrive at a close approximation to image B. A new location is specified for every source pixel in such a mapping, but it does not guarantee that all target image pixels will be covered with a pixel from the source image, and so (as previously discussed) "holes" may arise. The inverse map from A to B indicates, for each pixel in image B, the coordinates (or offsets) of corresponding pixels in image A. In using the inverse map, no holes occur, since every target pixel has a defined source pixel in the undeformed image. On the other hand, defining the source pixel coordinates for regions outside the on-screen boundaries of an object that does not fill the screen (like the teapot) may be problematic. In order to achieve reasonable results when interpolating inverse maps, the regions of the maps outside the boundaries of objects should contain coordinates or offsets extrapolated from the coordinates or offsets within object boundaries. The boundaries themselves, which delimit the visible portions of objects, can be represented separately, as mattes. A matte is an extra component of the image - along with the red, green, and blue components - which defines the opacity of the image at each point. The mattes for the source and target images in a moφh are indexed, deformed, and composited in the same way as the red, green, and blue components of the images.
It will be appreciated that the foregoing examples of the invention are illustrative only, and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics thereof. For example, while the use of linear interpolation has been described in the illustrated embodiments, other types of interpolation, such as quadratic or cubic, can also be employed to provide improved accuracy in the interpolated image. The disclosed embodiments, therefore, should not be considered to be restrictive in any sense. The scope of the invention is indicated by the appended claims, rather than the preceding description, and all variations which fall within the meaning and range of equivalents thereof are intended to be embraced therein.

Claims

What is claimed is:
1. A method for synthesizing views of a three- dimensional scene, comprising the steps of: storing a plurality of two-dimensional images of the scene from different respective source viewpoints; for each pair of images relating to adjacent viewpoints, determining and storing a map which identifies the correspondence of pixels in the two images; identifying a viewpoint between at least two of said source viewpoints; determining a coefficient relating to the location of said intermediate viewpoint relative to said source viewpoints; moφhing at least one of said images in accordance with said coefficient and the information contained in at least one of said maps to produce an interpolated image which defines the view from said intermediate viewpoint; and displaying the interpolated image.
2. The method of claim 1 wherein multiple images are moφhed, using multivariate interpolation, to produce a composite image that comprises said interpolated image.
3. The method of claim 1 wherein a pair of bidirectional offset maps are determined and stored for each pair of adjacent images.
4. The method of claim 1 wherein each map contains offset vectors which describe spatial displacement of image pixels in going from one image to another.
5. The method of claim 1, further including the step of decomposing said maps into blocks of pixels, in accordance with a compression threshold.
6. The method of claim 5, wherein the offset map identifies the correspondence of groups of pixels in the two images.
7. The method of claim 5, further including the step of sorting the blocks of pixels within a map in accordance with range information.
8. The method of claim 7 wherein said step of producing an interpolated image includes the steps of: copying pixel blocks from the source image to the destination image in said sorted order; identifying pixels in the destination image to which no pixel from the source image was copied; and for each identified pixel, filtering a parameter associated with at least one pixel in the neighborhood of said identified pixel, to compute a display value for the identified pixel.
9. The method of claim 8 wherein said parameter is the color of the adjacent pixel.
10. The method of claim 8 wherein said parameter is a displacement value for the adjacent pixel.
11. The method of claim 1, wherein each source viewpoint has an associated field of view defined by a central axis and edge boundaries, and wherein said source viewpoints are selected such that the angle between the edge boundary of one viewpoint and the central axis of an adjacent viewpoint does not exceed 90°.
12. A system for generating a sequence of related images corresponding to different views of a scene, comprising: means for storing an array of images which correspond to views of the scene from different respective source viewpoints; means for determining the correspondence between adjacent images in said array and storing maps which identify said correspondence; means for indicating a new viewpoint located between at least two of said source viewpoints; means for interpolating the information in at least one of said maps to produce an interpolated image corresponding to a view of the scene from said new viewpoint; and means for displaying the interpolated image.
13. The system of claim 12 further including means for decomposing said maps into blocks of pixels, wherein the pixels in each block have a common characteristic relating to the correspondence between images.
14. The system of claim 13 wherein said decomposing means includes means for determining the value of a parameter for each block which relates to distance from a viewpoint, and means for storing said blocks in an order determined by said parameter values.
15. The system of claim 14 wherein said interpolating means interpolates and displays said blocks in said stored order.
16. The system of claim 12 wherein said array is multi¬ dimensional, and said interpolated image is interpolated from at least two of said maps.
PCT/US1994/008467 1993-08-03 1994-07-26 Three-dimensional image synthesis using view interpolation WO1995004331A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
AU73743/94A AU7374394A (en) 1993-08-03 1994-07-26 Three-dimensional image synthesis using view interpolation

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US08/100,923 1993-08-03
US08/100,923 US5613048A (en) 1993-08-03 1993-08-03 Three-dimensional image synthesis using view interpolation

Publications (1)

Publication Number Publication Date
WO1995004331A1 true WO1995004331A1 (en) 1995-02-09

Family

ID=22282228

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US1994/008467 WO1995004331A1 (en) 1993-08-03 1994-07-26 Three-dimensional image synthesis using view interpolation

Country Status (3)

Country Link
US (1) US5613048A (en)
AU (1) AU7374394A (en)
WO (1) WO1995004331A1 (en)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0757335A2 (en) * 1995-08-02 1997-02-05 Nippon Hoso Kyokai 3D object graphics display device and method
EP0773516A1 (en) * 1995-11-13 1997-05-14 Atelier de Production Multimedia Virtual camera console
WO1998043208A2 (en) * 1997-03-21 1998-10-01 Newfire, Inc. Method and apparatus for graphics processing
US6426757B1 (en) 1996-03-04 2002-07-30 International Business Machines Corporation Method and apparatus for providing pseudo-3D rendering for virtual reality computer user interfaces
EP1289317A1 (en) * 2000-03-31 2003-03-05 Olympus Optical Co., Ltd. Method for posting three-dimensional image data and system for creating three-dimensional image
GB2462095A (en) * 2008-07-23 2010-01-27 Snell & Wilcox Ltd Processing of images to represent a transition in viewpoint
CN107270875A (en) * 2017-05-09 2017-10-20 南京航空航天大学 Visual signature three-dimensional rebuilding method under motion blur effects
US10284454B2 (en) 2007-11-30 2019-05-07 Activision Publishing, Inc. Automatic increasing of capacity of a virtual space in a virtual world
US10627983B2 (en) 2007-12-24 2020-04-21 Activision Publishing, Inc. Generating data for managing encounters in a virtual world environment
US11972086B2 (en) 2019-03-18 2024-04-30 Activision Publishing, Inc. Automatic increasing of capacity of a virtual space in a virtual world

Families Citing this family (144)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6005967A (en) * 1994-02-18 1999-12-21 Matushita Electric Industrial Co., Ltd. Picture synthesizing apparatus and method
US7843497B2 (en) 1994-05-31 2010-11-30 Conley Gregory J Array-camera motion picture device, and methods to produce new visual and aural effects
TW304254B (en) * 1994-07-08 1997-05-01 Hitachi Ltd
JPH0896165A (en) * 1994-09-29 1996-04-12 Fujitsu Ltd Method and device for image generation
US5990904A (en) * 1995-08-04 1999-11-23 Microsoft Corporation Method and system for merging pixel fragments in a graphics rendering system
US5864342A (en) * 1995-08-04 1999-01-26 Microsoft Corporation Method and system for rendering graphical objects to image chunks
US5852443A (en) * 1995-08-04 1998-12-22 Microsoft Corporation Method and system for memory decomposition in a graphics rendering system
US5949428A (en) * 1995-08-04 1999-09-07 Microsoft Corporation Method and apparatus for resolving pixel data in a graphics rendering system
US5870097A (en) * 1995-08-04 1999-02-09 Microsoft Corporation Method and system for improving shadowing in a graphics rendering system
US6009188A (en) * 1996-02-16 1999-12-28 Microsoft Corporation Method and system for digital plenoptic imaging
US6445807B1 (en) * 1996-03-22 2002-09-03 Canon Kabushiki Kaisha Image processing method and apparatus
US6529193B1 (en) * 1996-06-25 2003-03-04 Mental Images Gmbh & Co. Kg System and method for generating pixel values for pixels in an image using strictly deterministic methodologies for generating sample points
AU5896498A (en) 1996-12-06 1998-07-15 Stereographics Corporation Synthetic panoramagram
IL119928A (en) * 1996-12-29 2001-01-11 Univ Ramot Model-based view extrapolation for interactive virtual reality systems
US6097394A (en) * 1997-04-28 2000-08-01 Board Of Trustees, Leland Stanford, Jr. University Method and system for light field rendering
US5963213A (en) * 1997-05-07 1999-10-05 Olivr Corporation Ltd. Method and system for accelerating warping
US6094199A (en) * 1997-05-23 2000-07-25 University Of Washington 3D objects morphing employing skeletons indicating symmetric differences to define intermediate objects used in morphing
US5982375A (en) * 1997-06-20 1999-11-09 Sun Microsystems, Inc. Floating point processor for a three-dimensional graphics accelerator which includes single-pass stereo capability
EP0896301A1 (en) * 1997-08-06 1999-02-10 Victor Company Of Japan, Limited Stereoscopic image interpolating apparatus and method
JP3059956B2 (en) * 1998-03-19 2000-07-04 コナミ株式会社 IMAGE CREATING DEVICE, IMAGE CREATING METHOD, AND READABLE RECORDING MEDIUM CONTAINING IMAGE CREATING PROGRAM
US6317139B1 (en) 1998-03-25 2001-11-13 Lance Williams Method and apparatus for rendering 3-D surfaces from 2-D filtered silhouettes
DE19816795A1 (en) 1998-04-16 1999-10-21 Bosch Gmbh Robert Representations of objects in a bitmap format
US6285794B1 (en) * 1998-04-17 2001-09-04 Adobe Systems Incorporated Compression and editing of movies by multi-image morphing
US6201546B1 (en) * 1998-05-29 2001-03-13 Point Cloud, Inc. Systems and methods for generating three dimensional, textured models
US7071949B1 (en) * 1998-11-18 2006-07-04 Microsoft Corporation View dependent tiled textures
GB9901514D0 (en) 1999-01-25 1999-03-17 Koninkl Philips Electronics Nv Displaying images
JP3634677B2 (en) * 1999-02-19 2005-03-30 キヤノン株式会社 Image interpolation method, image processing method, image display method, image processing apparatus, image display apparatus, and computer program storage medium
US6906708B1 (en) * 1999-03-26 2005-06-14 Canon Kabushiki Kaisha Image processing method and apparatus, and storage medium
JP2001067463A (en) * 1999-06-22 2001-03-16 Nadeisu:Kk Device and method for generating facial picture from new viewpoint based on plural facial pictures different in viewpoint, its application device and recording medium
US7292261B1 (en) 1999-08-20 2007-11-06 Patrick Teo Virtual reality camera
US6847364B1 (en) * 1999-12-23 2005-01-25 Intel Corporation Methods and apparatus for creating three-dimensional motion illusion in a graphics processing system
AUPQ691100A0 (en) * 2000-04-14 2000-05-11 Lim, Dr Hong Lip Improvements to 3d graphics
US6573912B1 (en) 2000-11-07 2003-06-03 Zaxel Systems, Inc. Internet system for virtual telepresence
EP1209907A1 (en) * 2000-11-27 2002-05-29 Sony Service Centre (Europe) N.V. Display device
JP2002232908A (en) * 2000-11-28 2002-08-16 Monolith Co Ltd Image interpolation method and device
US6724417B1 (en) * 2000-11-29 2004-04-20 Applied Minds, Inc. Method and apparatus maintaining eye contact in video delivery systems using view morphing
WO2002050777A2 (en) * 2000-12-18 2002-06-27 Alan Sullivan Rasterization of three-dimensional images
JP2002259979A (en) * 2000-12-26 2002-09-13 Monolith Co Ltd Method and device for image interpolation, and method and device for image processing
KR100454070B1 (en) * 2001-01-31 2004-10-20 학교법인 중앙대학교 Method for Real-time Toon Rendering with Shadow using computer
US7133052B1 (en) * 2001-03-20 2006-11-07 Microsoft Corporation Morph map based simulated real-time rendering
US20020158972A1 (en) * 2001-04-25 2002-10-31 Stephany Thomas M. Method of image processing an animated figure
US7345783B2 (en) * 2001-10-31 2008-03-18 Canon Kabushiki Kaisha Image processing method and apparatus for generating sequential still images by extracting still images from moving image data, and printing apparatus
US6795090B2 (en) 2001-11-13 2004-09-21 Eastman Kodak Company Method and system for panoramic image morphing
FR2842641B1 (en) * 2002-07-19 2005-08-05 St Microelectronics Sa IMAGE DISPLAY ON A MATRIX SCREEN
JP3721155B2 (en) * 2002-10-30 2005-11-30 株式会社東芝 Information processing apparatus and display control method for the same
US7242400B2 (en) * 2002-11-13 2007-07-10 Ati Technologies Ulc Compression and decompression of data using plane equations
AU2002952874A0 (en) * 2002-11-25 2002-12-12 Dynamic Digital Depth Research Pty Ltd 3D image synthesis from depth encoded source view
KR100501907B1 (en) * 2002-12-23 2005-07-25 한국전자통신연구원 Method for Geometric Shape Morphing using Direction Map
US20040130578A1 (en) * 2003-01-08 2004-07-08 Douglas Charney Method and system for selection of menu items in a telephone menu screen interface
US20040222987A1 (en) * 2003-05-08 2004-11-11 Chang Nelson Liang An Multiframe image processing
JP3867143B2 (en) * 2003-06-25 2007-01-10 独立行政法人産業技術総合研究所 Three-dimensional microscope system and image display method
US7027081B2 (en) * 2003-12-21 2006-04-11 Kremen Stanley H System and apparatus for recording, transmitting, and projecting digital three-dimensional images
FR2864878B1 (en) * 2004-01-06 2006-04-14 Thomson Licensing Sa METHOD AND SYSTEM FOR DETERMINING THE MOVEMENT OF A PIXEL, AND RECORDING MEDIUM FOR IMPLEMENTING THE METHOD
US7817871B2 (en) * 2004-08-18 2010-10-19 Adobe Systems Incorporated Scaling of raster images without blurring of edges
JP2006058985A (en) * 2004-08-18 2006-03-02 Sony Corp Display control apparatus and method, and program
CA2511040A1 (en) * 2004-09-23 2006-03-23 The Governors Of The University Of Alberta Method and system for real time image rendering
EP1820159A1 (en) 2004-11-12 2007-08-22 MOK3, Inc. Method for inter-scene transitions
US7306341B2 (en) * 2005-02-28 2007-12-11 Hewlett-Packard Development Company, L.P. Multi-projector geometric calibration
US7860344B1 (en) * 2005-05-06 2010-12-28 Stochastech Corporation Tracking apparatus and methods using image processing noise reduction
US7542038B2 (en) * 2005-06-30 2009-06-02 Intel Corporation Method and system for pivot-point based adaptive soft shadow estimation
US20070046924A1 (en) * 2005-08-30 2007-03-01 Chang Nelson L A Projecting light patterns encoding correspondence information
US7702131B2 (en) * 2005-10-13 2010-04-20 Fujifilm Corporation Segmenting images and simulating motion blur using an image sequence
EP1951115A4 (en) * 2005-11-11 2013-10-30 Randall L Barbour Functional imaging of autoregulation
US20070153023A1 (en) * 2006-01-04 2007-07-05 Computer Associates Think, Inc. System and method of temporal anti-aliasing
TW200734618A (en) * 2006-03-06 2007-09-16 Benq Corp Bulb sorting device and sorting method thereof
JP4839128B2 (en) * 2006-05-10 2011-12-21 矢崎総業株式会社 Graphic meter display device
US7768527B2 (en) * 2006-05-31 2010-08-03 Beihang University Hardware-in-the-loop simulation system and method for computer vision
US7573475B2 (en) * 2006-06-01 2009-08-11 Industrial Light & Magic 2D to 3D image conversion
US7573489B2 (en) * 2006-06-01 2009-08-11 Industrial Light & Magic Infilling for 2D to 3D image conversion
AU2007319441A1 (en) 2006-11-13 2008-05-22 Everyscape, Inc. Method for scripting inter-scene transitions
US8159494B2 (en) * 2006-12-04 2012-04-17 Electronics And Telecommunications Research Institute System and method for designing curves
US7839422B2 (en) * 2006-12-13 2010-11-23 Adobe Systems Incorporated Gradient-domain compositing
US10063848B2 (en) * 2007-08-24 2018-08-28 John G. Posa Perspective altering display system
US8180167B2 (en) * 2008-07-16 2012-05-15 Seiko Epson Corporation Model-based error resilience in data communication
US20100066731A1 (en) * 2008-09-16 2010-03-18 James Calvin Vecore Configurator Process and System
US8248410B2 (en) * 2008-12-09 2012-08-21 Seiko Epson Corporation Synthesizing detailed depth maps from images
US8194097B2 (en) * 2008-12-12 2012-06-05 Seiko Epson Corporation Virtual masking using rigid parametric modeling
JP5182229B2 (en) * 2009-06-02 2013-04-17 ソニー株式会社 Image processing apparatus, image processing method, and program
US20110001791A1 (en) * 2009-07-02 2011-01-06 Emaze Imaging Techonolgies Ltd. Method and system for generating and displaying a three-dimensional model of physical objects
US8436893B2 (en) 2009-07-31 2013-05-07 3Dmedia Corporation Methods, systems, and computer-readable storage media for selecting image capture positions to generate three-dimensional (3D) images
US9380292B2 (en) 2009-07-31 2016-06-28 3Dmedia Corporation Methods, systems, and computer-readable storage media for generating three-dimensional (3D) images of a scene
US8508580B2 (en) 2009-07-31 2013-08-13 3Dmedia Corporation Methods, systems, and computer-readable storage media for creating three-dimensional (3D) images of a scene
SG176327A1 (en) 2010-05-20 2011-12-29 Sony Corp A system and method of image processing
US8817017B2 (en) 2010-06-01 2014-08-26 Vladimir Vaganov 3D digital painting
US10922870B2 (en) * 2010-06-01 2021-02-16 Vladimir Vaganov 3D digital painting
US10217264B2 (en) * 2010-06-01 2019-02-26 Vladimir Vaganov 3D digital painting
US9734622B2 (en) 2010-06-01 2017-08-15 Vladimir Vaganov 3D digital painting
US9440484B2 (en) 2010-06-01 2016-09-13 Vladimir Vaganov 3D digital painting
US8619124B2 (en) * 2010-10-14 2013-12-31 Industrial Technology Research Institute Video data processing systems and methods
US9185388B2 (en) 2010-11-03 2015-11-10 3Dmedia Corporation Methods, systems, and computer program products for creating three-dimensional video sequences
WO2012092246A2 (en) 2010-12-27 2012-07-05 3Dmedia Corporation Methods, systems, and computer-readable storage media for identifying a rough depth map in a scene and for determining a stereo-base distance for three-dimensional (3d) content creation
US10200671B2 (en) 2010-12-27 2019-02-05 3Dmedia Corporation Primary and auxiliary image capture devices for image processing and related methods
US8274552B2 (en) 2010-12-27 2012-09-25 3Dmedia Corporation Primary and auxiliary image capture devices for image processing and related methods
EP2547111B1 (en) 2011-07-12 2017-07-19 Samsung Electronics Co., Ltd. Method and apparatus for processing multi-view image using hole rendering
US9189891B2 (en) * 2011-08-16 2015-11-17 Google Inc. Systems and methods for navigating a camera
US8274524B1 (en) 2011-09-28 2012-09-25 Google Inc. Map rendering using interpolation of style parameters across zoom levels
US9613453B2 (en) * 2011-10-04 2017-04-04 Google Inc. Systems and method for performing a three pass rendering of images
US8243102B1 (en) 2011-10-12 2012-08-14 Google Inc. Derivative-based selection of zones for banded map display
BE1019941A3 (en) 2012-06-05 2013-02-05 Tait Technologies Bvba DEVICE FOR VIEWING THREE-DIMENSIONAL IMAGES, CREATION OF THREE-DIMENSIONAL IMAGES, AND METHOD FOR CREATING THREE-DIMENSIONAL IMAGES.
US20150334276A1 (en) * 2012-12-31 2015-11-19 Given Imaging Ltd. System and method for displaying an image stream
JP6187811B2 (en) * 2013-09-09 2017-08-30 ソニー株式会社 Image processing apparatus, image processing method, and program
KR20150033162A (en) * 2013-09-23 2015-04-01 삼성전자주식회사 Compositor and system-on-chip having the same, and driving method thereof
KR102158390B1 (en) 2013-10-22 2020-09-22 삼성전자주식회사 Method and apparatus for image processing
US10244223B2 (en) 2014-01-10 2019-03-26 Ostendo Technologies, Inc. Methods for full parallax compressed light field 3D imaging systems
CN103871109B (en) * 2014-04-03 2017-02-22 深圳市德赛微电子技术有限公司 Virtual reality system free viewpoint switching method
GB2524960B (en) * 2014-04-04 2019-05-15 Imagineer Systems Ltd Processing of digital motion images
US10930038B2 (en) 2014-06-10 2021-02-23 Lab Of Misfits Ar, Inc. Dynamic location based digital element
US10026226B1 (en) * 2014-06-10 2018-07-17 Ripple Inc Rendering an augmented reality object
US10008027B1 (en) 2014-10-20 2018-06-26 Henry Harlyn Baker Techniques for determining a three-dimensional representation of a surface of an object from a set of images
US10275935B2 (en) 2014-10-31 2019-04-30 Fyusion, Inc. System and method for infinite synthetic image generation from multi-directional structured image array
US10176592B2 (en) 2014-10-31 2019-01-08 Fyusion, Inc. Multi-directional structured image array capture on a 2D graph
US10262426B2 (en) 2014-10-31 2019-04-16 Fyusion, Inc. System and method for infinite smoothing of image sequences
US9940541B2 (en) 2015-07-15 2018-04-10 Fyusion, Inc. Artificially rendering images using interpolation of tracked control points
US10726593B2 (en) 2015-09-22 2020-07-28 Fyusion, Inc. Artificially rendering images using viewpoint interpolation and extrapolation
JP6866299B2 (en) 2015-04-23 2021-04-28 オステンド・テクノロジーズ・インコーポレーテッド Methods and equipment for omnidirectional parallax light field display systems
WO2016172385A1 (en) 2015-04-23 2016-10-27 Ostendo Technologies, Inc. Methods for full parallax compressed light field synthesis utilizing depth information
JP6250592B2 (en) * 2015-06-02 2017-12-20 株式会社ソニー・インタラクティブエンタテインメント Head mounted display, information processing apparatus, display control method, and program
US11095869B2 (en) 2015-09-22 2021-08-17 Fyusion, Inc. System and method for generating combined embedded multi-view interactive digital media representations
US10222932B2 (en) 2015-07-15 2019-03-05 Fyusion, Inc. Virtual reality environment based manipulation of multilayered multi-view interactive digital media representations
US10147211B2 (en) * 2015-07-15 2018-12-04 Fyusion, Inc. Artificially rendering images using viewpoint interpolation and extrapolation
US10852902B2 (en) 2015-07-15 2020-12-01 Fyusion, Inc. Automatic tagging of objects on a multi-view interactive digital media representation of a dynamic entity
US10698558B2 (en) 2015-07-15 2020-06-30 Fyusion, Inc. Automatic tagging of objects on a multi-view interactive digital media representation of a dynamic entity
US10750161B2 (en) 2015-07-15 2020-08-18 Fyusion, Inc. Multi-view interactive digital media representation lock screen
US10242474B2 (en) 2015-07-15 2019-03-26 Fyusion, Inc. Artificially rendering images using viewpoint interpolation and extrapolation
US11006095B2 (en) 2015-07-15 2021-05-11 Fyusion, Inc. Drone based capture of a multi-view interactive digital media
US11783864B2 (en) 2015-09-22 2023-10-10 Fyusion, Inc. Integration of audio into a multi-view interactive digital media representation
US10448030B2 (en) 2015-11-16 2019-10-15 Ostendo Technologies, Inc. Content adaptive light field compression
US10453431B2 (en) 2016-04-28 2019-10-22 Ostendo Technologies, Inc. Integrated near-far light field display systems
US10403033B2 (en) * 2016-07-12 2019-09-03 Microsoft Technology Licensing, Llc Preserving scene lighting effects across viewing perspectives
US11202017B2 (en) 2016-10-06 2021-12-14 Fyusion, Inc. Live style transfer on a mobile device
US10430994B1 (en) 2016-11-07 2019-10-01 Henry Harlyn Baker Techniques for determining a three-dimensional textured representation of a surface of an object from a set of images with varying formats
US10437879B2 (en) 2017-01-18 2019-10-08 Fyusion, Inc. Visual search using multi-view interactive digital media representations
US10313651B2 (en) 2017-05-22 2019-06-04 Fyusion, Inc. Snapshots at predefined intervals or angles
US20180350038A1 (en) 2017-06-02 2018-12-06 Ostendo Technologies, Inc. Methods and Systems for Light Field Compression With Residuals
US11051039B2 (en) 2017-06-02 2021-06-29 Ostendo Technologies, Inc. Methods for full parallax light field compression
US11069147B2 (en) 2017-06-26 2021-07-20 Fyusion, Inc. Modification of multi-view interactive digital media representation
US10931956B2 (en) 2018-04-12 2021-02-23 Ostendo Technologies, Inc. Methods for MR-DIBR disparity map merging and disparity threshold determination
US10592747B2 (en) 2018-04-26 2020-03-17 Fyusion, Inc. Method and apparatus for 3-D auto tagging
US11172222B2 (en) 2018-06-26 2021-11-09 Ostendo Technologies, Inc. Random access in encoded full parallax light field images
RU2690757C1 (en) 2018-08-21 2019-06-05 Самсунг Электроникс Ко., Лтд. System for synthesis of intermediate types of light field and method of its operation
US11087549B2 (en) 2018-10-15 2021-08-10 University Of Maryland, College Park Methods and apparatuses for dynamic navigable 360 degree environments
CN109493406B (en) * 2018-11-02 2022-11-11 四川大学 Fast percentage approaching soft shadow drawing method
US20230063215A1 (en) * 2020-01-23 2023-03-02 Sony Group Corporation Information processing apparatus, information processing method, and program
CN111951369B (en) * 2020-09-01 2023-05-23 网易(杭州)网络有限公司 Detail texture processing method and device

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4935879A (en) * 1987-08-05 1990-06-19 Daikin Industries, Ltd. Texture mapping apparatus and method
US5204944A (en) * 1989-07-28 1993-04-20 The Trustees Of Columbia University In The City Of New York Separable image warping methods and systems using spatial lookup tables
US5255352A (en) * 1989-08-03 1993-10-19 Computer Design, Inc. Mapping of two-dimensional surface detail on three-dimensional surfaces
US5193126A (en) * 1990-01-25 1993-03-09 Toyo Ink Mfg. Co., Ltd. Image deformation method and apparatus using coordinate transformation
US5224208A (en) * 1990-03-16 1993-06-29 Hewlett-Packard Company Gradient calculation for texture mapping
US5230039A (en) * 1991-02-19 1993-07-20 Silicon Graphics, Inc. Texture range controls for improved texture mapping
US5307450A (en) * 1991-02-19 1994-04-26 Silicon Graphics, Inc. Z-subdivision for improved texture mapping
US5187754A (en) * 1991-04-30 1993-02-16 General Electric Company Forming, with the aid of an overview image, a composite image from a mosaic of images
US5416899A (en) * 1992-01-13 1995-05-16 Massachusetts Institute Of Technology Memory based method and apparatus for computer graphics
US5325475A (en) * 1992-09-09 1994-06-28 Massachusetts Institute Of Technology Computer method and apparatus for matching between line drawings

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
DEVARAJAN AND HOOKS: "LOW ALTITUDE HIGH SPEED FLIGHT SIMULATION USING VIDEO DISC TECHNOLOGY", LTV AEROSPACE AND DEFENSE COMPANY, 25 July 1985 (1985-07-25), pages 54 - 65 *
KAUL AND ROSSIGNAC: "SOLID-INTERPOLATING DEFORMATIONS: CONSTRUCTION AND ANIMATION OF PIPS", COMPUTER & GRAPHICS, vol. 16, no. 1, 1992, GREAT BRITAIN, pages 107 - 115 *

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0757335A2 (en) * 1995-08-02 1997-02-05 Nippon Hoso Kyokai 3D object graphics display device and method
EP0757335A3 (en) * 1995-08-02 1999-01-07 Nippon Hoso Kyokai 3D object graphics display device and method
EP0773516A1 (en) * 1995-11-13 1997-05-14 Atelier de Production Multimedia Virtual camera console
FR2741225A1 (en) * 1995-11-13 1997-05-16 Production Multimedia Apm Atel VIRTUAL CAMERA CONSOLE
US6426757B1 (en) 1996-03-04 2002-07-30 International Business Machines Corporation Method and apparatus for providing pseudo-3D rendering for virtual reality computer user interfaces
WO1998043208A2 (en) * 1997-03-21 1998-10-01 Newfire, Inc. Method and apparatus for graphics processing
WO1998043208A3 (en) * 1997-03-21 1999-01-14 Newfire Inc Method and apparatus for graphics processing
EP1289317A4 (en) * 2000-03-31 2007-03-21 Olympus Optical Co Method for posting three-dimensional image data and system for creating three-dimensional image
EP1289317A1 (en) * 2000-03-31 2003-03-05 Olympus Optical Co., Ltd. Method for posting three-dimensional image data and system for creating three-dimensional image
US10284454B2 (en) 2007-11-30 2019-05-07 Activision Publishing, Inc. Automatic increasing of capacity of a virtual space in a virtual world
US10627983B2 (en) 2007-12-24 2020-04-21 Activision Publishing, Inc. Generating data for managing encounters in a virtual world environment
GB2462095A (en) * 2008-07-23 2010-01-27 Snell & Wilcox Ltd Processing of images to represent a transition in viewpoint
US8515205B2 (en) 2008-07-23 2013-08-20 Snell Limited Processing of images to represent a transition in viewpoint
CN107270875A (en) * 2017-05-09 2017-10-20 南京航空航天大学 Visual signature three-dimensional rebuilding method under motion blur effects
CN107270875B (en) * 2017-05-09 2020-04-24 南京航空航天大学 Visual feature three-dimensional reconstruction method under motion blur effect
US11972086B2 (en) 2019-03-18 2024-04-30 Activision Publishing, Inc. Automatic increasing of capacity of a virtual space in a virtual world

Also Published As

Publication number Publication date
AU7374394A (en) 1995-02-28
US5613048A (en) 1997-03-18

Similar Documents

Publication Publication Date Title
US5613048A (en) Three-dimensional image synthesis using view interpolation
Chen et al. View interpolation for image synthesis
CN111508052B (en) Rendering method and device of three-dimensional grid body
Debevec et al. Efficient view-dependent image-based rendering with projective texture-mapping
US6529207B1 (en) Identifying silhouette edges of objects to apply anti-aliasing
Aliaga Visualization of complex models using dynamic texture-based simplification
CA2232757C (en) Real-time image rendering with layered depth images
Shum et al. Pop-up light field: An interactive image-based modeling and rendering system
EP0638875B1 (en) A 3-dimensional animation generating apparatus and a method for generating a 3-dimensional animation
US20050219249A1 (en) Integrating particle rendering and three-dimensional geometry rendering
US6326972B1 (en) 3D stroke-based character modeling suitable for efficiently rendering large crowds
JPH0778267A (en) Method for display of shadow and computer-controlled display system
JPH0757117A (en) Forming method of index to texture map and computer control display system
GB2259432A (en) Three dimensional graphics processing
GB2392072A (en) Generating shadow image data of a 3D object
GB2256567A (en) Modelling system for imaging three-dimensional models
JP2004522224A (en) Synthetic rendering of 3D graphical objects
US20070120858A1 (en) Generation of motion blur
Bonatto et al. Real-time depth video-based rendering for 6-DoF HMD navigation and light field displays
Shelley et al. Path specification and path coherence
US5793372A (en) Methods and apparatus for rapidly rendering photo-realistic surfaces on 3-dimensional wire frames automatically using user defined points
WO2023004559A1 (en) Editable free-viewpoint video using a layered neural representation
EP0656609B1 (en) Image processing
US6690369B1 (en) Hardware-accelerated photoreal rendering
US6906729B1 (en) System and method for antialiasing objects

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AM AT AU BB BG BR BY CA CH CN CZ DE DK ES FI GB GE HU JP KE KG KP KR KZ LK LT LU LV MD MG MN MW NL NO NZ PL PT RO RU SD SE SI SK TJ TT UA US UZ VN

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): KE MW SD AT BE CH DE DK ES FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN ML MR NE SN TD TG

DFPE Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)
121 Ep: the epo has been informed by wipo that ep was designated in this application
REG Reference to national code

Ref country code: DE

Ref legal event code: 8642

122 Ep: pct application non-entry in european phase
NENP Non-entry into the national phase

Ref country code: CA