Imparting Three-Dimensional Characteristics in a Two-Dimensional Space

Abstract

Processes and techniques for imparting three-dimensional visual characteristics to images in a two-dimensional space are described. In one implementation, a graphical image is received in a two-dimensional space (e.g., a user interface on a computing device). A shape palette is presented to a user, the shape palette comprising a visual representation of three-dimensional visual information. Based on user markup of the shape palette, three-dimensional visual information is extracted from the shape palette and correlated with the graphical image. The three-dimensional visual information is processed to render the graphical image with three-dimensional visual characteristics.

Description

BACKGROUND

Today's computer user has access to vast data processing and storage resources. These resources enable users to accomplish computing tasks that were impossible using previous generations of computers, particularly in the area of computer graphics. Modern computer graphics tools take advantage of these resources to create and display visually complex and highly detailed computer-generated images. The challenge then becomes to find user-friendly ways to input the large quantities of data required to define computer-generated images.

Most current computer-graphics tools enable users to modify or create graphics that have the appearance of depth, or appear to be three-dimensional. These tools typically create the appearance of three-dimensionality in a computer image by creating a mathematical representation of a three-dimensional object. Inputting the data needed to create the mathematical representation can be a very time consuming and tedious process. For example, most computer graphics tools require a user to specify three-dimensional information in multiple different image views and thus present a significant burden to users.

SUMMARY

This summary is provided to introduce techniques and processes for graphics processing, which are further described below in the Detailed Description. This summary is not intended to identify essential features of the claimed subject matter, nor is it intended for use in determining the scope of the claimed subject matter.

This disclosure is directed to processes and techniques for imparting three-dimensional characteristics to images in a two-dimensional space. In one implementation, a user provides markup in an image canvas interface and provides corresponding markup in a shape palette. The shape palette presents a visual representation of three-dimensional information. Based on the user markup three-dimensional information is extracted from the shape palette and correlated with the user markup in the image canvas. The correlated three-dimensional information and user markup from the image canvas are processed to render an image with three-dimensional characteristics.

BRIEF DESCRIPTION OF THE DRAWINGS

The detailed description is set forth with reference to the accompanying figures. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. The use of the same reference numbers in different figures indicates similar or identical items.

FIG. 1 illustrates one example of a user interface for rendering an image with three-dimensional characteristics based on user markup of an image canvas and user markup of a shape palette.

FIG. 2 illustrates one example of correlation between an image canvas and a shape palette to produce an image with three-dimensional characteristics.

FIG. 3 illustrates one example of a shape palette.

FIG. 4 illustrates one example of a correlation between an image canvas and a shape palette to produce an image with three-dimensional characteristics.

FIG. 5 is a flow diagram of an illustrative process of rendering an image with three-dimensional characteristics based on user markup of an image canvas and user markup of a shape palette.

FIG. 6 is a flow diagram detailing certain aspects of an illustrative process of rendering an image with three-dimensional characteristics based on user markup of an image canvas and user markup of a shape palette.

FIG. 7 illustrates one example of an apparatus for rendering an image with three-dimensional characteristics based on user markup of an image canvas and user markup of a shape palette.

DETAILED DESCRIPTION

Described herein are processes and techniques for imparting visual characteristics of three-dimensionality to graphic images in a two-dimensional space. The processes and techniques recognize that humans have the ability to infer three-dimensional (3D) structure from two-dimensional (2D) images. This ability applies to a variety of image formats, such as photographs, paintings, sketches, and line art. Often only a few 2D strokes are necessary to express the 3D shape of an object. An observer can often make an analogy between a stroke or mark in a 2D space and a corresponding 3D aspect or surface. The challenge in a computer graphics interface is to enable a user to use this ability to infer 3D structure to specify 3D information that can be interpreted by a computer graphics tool.

To enable a user to quickly specify 3D information in a 2D space (e.g., a computer graphics interface), the shape palette is introduced. In one example, a shape palette is a 2D image of a 3D object that stores salient 3D geometric information that can be extracted through user markup of the shape palette. A user provides a 2D primitive (e.g., a line, stroke or other shape) in a space on a user interface and then draws a corresponding primitive on the shape palette, 3D information is extracted from the shape palette based on the user markup and processed. Processing the 3D information can include copying, or warping, the 3D information to the 2D user interface. Multiple primitives can be marked up using a shape palette and rendered and combined to produce an overall image. From the transfer of a few sparse primitives, a 3D model surface can be extracted based on the 3D information transferred from the shape palette to the user interface. The shape palette can be marked-up in a single view (i.e., the shape palette need not be rotated or otherwise manipulated) to extract the desired 3D information. This is not intended to be limiting, however, and some implementations may utilize multiple shape palette views. Thus, the shape palette serves as an intuitive interface whereby a user can specify 3D visual aspects that the user wishes to impart to a 2D primitive or other visual image.

One approach to implementing the shape palette is to configure the shape palette as a 3D normal map. The normal map can be presented, for example, as a Lambertian-shaded image, such as a sphere. In one such example, the transfer of 3D information from the shape palette to a primitive involves the transfer of normal information from the shape palette to the primitive. Thus, a shape palette can comprise a plurality of pixels wherein each of the plurality of pixels is associated with a normal value. With a minimum amount of shape palette markup, a user can generate a sparse normal map. The sparse normal map is then processed to produce a dense normal map that can be further processed to produce a 3D surface or height image.

A shape palette can be generated using an orthographic projection of a 3D shape model and can store the inner product of a synthetic light and the 3D normal at each point in the projection (i.e., the shape palette utilizes Lambertian shading, for example). Shape palettes can consist of any suitable 3D model, such as sphere, a cone, a torus, and so on. Suitable ways of processing the 3D normal information from a shape palette to produce a dense normal field are discussed below.

Illustrative User Interface

FIG. 1 shows at 100 a graphics tool interface 102 that can implement the described processes and techniques. Graphics tool interface 102 is presented for purposes of example only, and the described processes and techniques can be implemented using different types and arrangements of user interfaces. Graphics tool interface 102 can be generated and/or displayed using any suitable device, including a desktop computer, a laptop, a personal digital assistant (PDA), a cellular phone, and so on. In this context, graphics tool interface 102 exists in a 2D space (i.e., the tool is displayed on a 2D interface, such as a computer monitor).

Graphics tool interface 102 includes image canvas 104. Image canvas 104 enables a user of graphics tool interface 102 to provide shapes and/or markup to which the user wishes to impart 3D visual characteristics. Also included in graphics tool interface 102 is shape palette 106. Shape palette 106 is configured to receive user markup, and based on the user markup, to impart 3D information to the shapes and/or markup indicated in image canvas 104. Rendering canvas 108 displays an image that is rendered based on the shapes and/or markup from image canvas 104 and the user markup provided in shape palette 106.

In operation, a user provides an image and/or markup in image canvas 104. Markup refers generally to providing strokes, lines, curves or other graphic input, such as by using a freehand tool as part of a graphics application. In one example, a user can import an image file (e.g., in JPEG, bitmap, or any other suitable graphics format) to the image canvas and use the image as a stencil over which markup can be provided in the image canvas. As the user provides markup in the image canvas, the user provides corresponding markup in the shape palette. For example, a user can make a stroke in image canvas 104 and make a corresponding stroke in shape palette 106. The graphics tool will automatically correlate the two strokes and impart the 3D information indicated by the shape palette stroke to the corresponding stroke in image canvas 104. However, as discussed below, other methods may be used to correlate shape palette markup with image canvas markup. This markup correlation is discussed in more detail below. Based on the correlation between the image canvas markup and the shape palette markup, an image is rendered in rendering canvas 108. The image rendered in rendering canvas 108 includes 3D visual characteristics extracted from shape palette 106 (e.g., Lambertian shading).

FIG. 2 shows, at 200, one example of rendering an image with 3D characteristics based on correlation between markup in an image canvas and markup in a shape palette. FIG. 2 is discussed with reference to graphics tool interface 102.

At 202, an image is illustrated that is marked up in an image canvas. In this example, an image file is imported into the image canvas and the user provides markup over the image. This is not required, however, and a user can provide markup to the image canvas without importing an image. In other examples, a user can import a graphic image and provide shape palette markup to impart 3D characteristics to the graphic image by correlating lines and other shapes from the graphic image with the shape palette markup. User markup of the image canvas is illustrated at 204 through 220. The user markup of the image canvas can be classified in two different types of user markup. A first type of user markup is silhouette markup. Silhouette markup denotes regions in the image canvas where the 3D information is tangent to the markup primitive and is consider to lie within the plane of the canvas. For example, if a 2D straight line is drawn as silhouette markup, then the 3D directions that will be associated with that straight-line will be directions perpendicular to that line. This means the 3D information can be computed directly from the 2D image canvas markup, therefore corresponding markup on the shape palette is not necessary. In this example, markups 204 and 206 are silhouette markups and therefore do not require corresponding 3D markup in the shape palette. This is not intended to be limiting, however, and other examples may utilize silhouette markup in both the image canvas and the shape palette.

A second type of user markup is non-silhouette markup. Non-silhouette markup has corresponding markup in the image palette and enables a user to impart 3D visual characteristics to a rendered image. In this example, markups 208 through 220 are non-silhouette markups and have corresponding markups in the shape palette.

Shape palettes are shown at 222 and 224. While these are shown as two separate shape palettes, the images can be displayed together (as shown in FIG. 1). Shape palette 222 represents a Lambertian-shaded image of a convex sphere. Thus, shape palette 222 can be visualized as 3D information extending outward from the plane of the 2D interface. Shape palette 224 represents a Lambertian-shaded image of a concave sphere. Thus, shape palette 224 can be visualized as 3D information extending inward from the plane of the 2D interface (i.e., away from the user's view).

As discussed above, to impart 3D visual characteristics to an image, a user may mark up the image canvas and make corresponding markup to the shape palette. In this example, the user marks up image 202 and makes corresponding markup to shape palettes 222 and 224. For example, the user provides markup 208 to image 202 and provides corresponding markup 226 to shape palette 222. The user then provides markup 210 to image 202 and provides corresponding markup 228 to shape palette 224, and so on for the remaining image and shape palette markups. Thus, in this example, in addition to the previously discussed correspondence, image markup 212 corresponds to shape palette markup 230, image markup 214 corresponds to shape palette markup 232, image markup 216 corresponds to shape palette markup 234, image markup 218 corresponds to shape palette markup 236, and image markup 220 corresponds to shape palette markup 238. In this example, image markups 204 and 206 are silhouette markups and do not require shape palette markup.

Once a user has provided markup to an image canvas (e.g., markup of image 202) and corresponding markup to a shape palette (e.g., shape palettes 222 and 224), a rendered image 240 can be generated using the 3D information extracted from the shape palette(s). In this example, rendered image 240 is rendered based on the markup of image 202 and the corresponding markup of shape palettes 222 and 224.

Illustrative Shape Palette

FIG. 3 shows at 300 an illustrative shape palette 106, introduced in FIG. 1. Shape palette 106 enables a user to choose an image or images to use as reference images to impart 3D information to markup in an image canvas. Shape palette 106 can be any suitable image that can be used to extract 3D geometric information. As shown in FIG. 3, examples of shape palettes include a sphere, a cylinder, a cone, a face, and so on. The shape palettes can be Lambertian-shaded to indicate 3D information. The shape palette may also include a drop-down menu or other user interface to allow a user to select the desired shape palette. This example is not intended to be limiting and any suitable type and arrangement of shape palette(s) can be utilized.

Illustrative Processes

Illustrative processes are described in this section with additional reference to FIGS. 1-3. The processes are described in the context of graphics tool interface 102 shown above for convenience only, and the processes are broadly applicable to other interfaces and systems.

The illustrative processes may be described in the general context of computer executable instructions and are illustrated as collections of blocks in logical flowcharts, which represent sequences of operations that can be implemented in hardware, software, or a combination thereof. Generally, computer executable instructions can include routines, programs, objects, components, data structures, procedures, modules, functions, and the like that perform particular functions or implement particular abstract data types. The processes may also be practiced in a distributed computing environment where functions are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, computer executable instructions may be located in both local and remote computer storage media, including memory storage devices.

FIG. 4 illustrates at 400 a process for imparting 3D characteristics to an image in a 2D space. At 402, image 404 is imported to an image canvas. At 406, the image canvas is marked up using image 404 as a stencil. At 408, a shape palette 410 is marked up with markup that corresponds to the markup of image 404 at 406. The illustrated shape palette includes a sphere and a face. The face shape palette contains typical 3D aspects of facial features, such as eye sockets, a nose, a mouth, and so on. The “face palette” can be used to mark up a more detailed facial image, such as image 404, to give the facial image 3D characteristics.

Markup 412 of image 404 is a silhouette markup and, in this example, does not require corresponding 3D markup in shape palette 410. Other markups are non-silhouette markups and thus have corresponding markups in shape palette 410. The correspondence of image 404 markup and shape palette 410 markup is as follows: Image markup 414 corresponds to shape palette markup 416; Image markup 418 corresponds to shape palette markup 420; Image markup 422 corresponds to shape palette markup 424; Image markup 426 corresponds to shape palette markup 428; and Image markup 430 corresponds to shape palette markup 432. Based on the markup correspondence, 3D information is extracted from the shape palettes and imparted to the image markups, at 434, to render an image with the 3D characteristics. At 436, the rendered image is used to relight image 404 and create image 438 that displays 3D characteristics. Thus, the discussed techniques can be used to impart 3D characteristics to 2D images.

FIG. 5 illustrates, at 500, an exemplary process for imparting 3D visual characteristics to an image based on user markup. At 502, a user provides markup in an image canvas. In one example, an image file can be loaded and an image from the image file displayed in the image canvas. In some examples, act 502 can include tracing, importing, and/or otherwise receiving user markup in an image canvas. At 504, the user marks up a shape palette to correspond with the image canvas markup. Thus, the user markup of the image canvas and the shape palette can be said to indicate data points within the image canvas and the shape palette. At 506, the shape palette markup is correlated with the corresponding image canvas markup. The image canvas markup and the shape palette markup can be provided in a “tit-for-tat” fashion such that after a user provides a particular stroke, line or image in the image canvas, the user then provides a corresponding markup in the shape palette. Other implementations may utilize different ways of correlating image canvas markup with shape palette markup, such as color-based correlation. For example, all markup of a specific color in the image canvas may be correlated with markup of the same or similar color in the shape palette. In other examples, the markup in the image canvas and the shape palette may not be provided in a “tit-for-tat” fashion, and the image canvas may be partially or completely marked up before the shape palette receives user markup. In that case, strokes in the image canvas may subsequently be correlated with strokes in the shape palette.

At 508, 3D information is extracted from the shape palette based on the user markup of the shape palette. The 3D information can include geometric information, such as normal values (e.g., in terms of x, y and z coordinates) and pixel values (e.g., brightness) for each point indicated in the user markup of the shape palette. At 510, the 3D information is processed to produce data that can generate an image with 3D characteristics. In some examples, normal values extracted from the shape palette are warped to the image canvas markup using geometric techniques such as thin-plate splines. At 512, an image with 3D characteristics is rendered using the data.

FIG. 6 illustrates more detailed aspects of one illustrative implementation of act 510, discussed above with reference to FIG. 5. The process shown at 510 is one process for generating an image with 3D characteristics using a sparse set of normals indicated by user markup of a shape palette. At 600, a set of normals and a set of pixels indicated by user markup of a shape palette are received. At 602, the set of normals and the set of pixels are processed to produce a dense normal map that can then be used to render an image with 3D characteristics. For purposes of example, one method of processing the set of pixels and the set of sparse normals is presented below.

A normal represents a 3D direction in Euclidean space and can be expressed as n=[x,y,z]^T. A unit normal is a normal whose x, y, and z components are of length one, i.e. 1=√{square root over (x²+y²+z²)}. An alternative representation of a normal may be denoted as

$n={\frac{1}{\sqrt{p^2+q^2+1}}\left[-p-q1\right]}^T,\mathrm{where}$ $p=-\frac{x}{z}$ $\mathrm{and}$ $p=-\frac{y}{z}$

associated with a unit normal [x,y,z]^T. Consider now we have an image, where each pixel is denoted by i, and we wish to have a normal associate with that pixel. We can denote each normal in the image

$n_i={\frac{1}{\sqrt{p_i^2+q_i^2+1}}\left[-p_i-q_i1\right]}^T.$

The shape palette can be considered as an image that has a normal at every pixel. The aforementioned markup procedure provides a way for the user to transfer the normals from the shape palette to the image canvas. In some cases, this transfer via markup is very spare, and the goal is to compute a dense normal image based on the sparse markup. The following illustrative method shows one way to estimate the p_i component of the normal n_i for all pixels in an image rendered with 3D characteristics. The same method can be used to estimate q_i.

Consider an image that is set of normal values, G={iε1 . . . N}. G is a set of p's and N is the total number of pixels of the canvas. G can be estimated using the sparse set of normal values generated by user markup of the shape palette. The set of normal values generated by user markup of the shape palette is represented by the observation set O={{tilde over (p)}_k|kεS}, where {tilde over (p)}_k is known and S is the set of corresponding pixel locations. G can be obtained by solving the following associated energy function:

E(G)=log(P(O|G))+log(P(G))   (1)

where (P(O|G)) is the likelihood and (P(G)) is the prior. The likelihood can be defined as:

$\begin{array}{cc}P\left(OG\right)=\prod _{k\in S}\exp \left(-\frac{{p_k-{\tilde{p}}_k}^2}{2{\sigma }_1^2}\right)& \left(2\right)\end{array}$

which measures the faithfulness of the output normal map to the input O; the prior P(G) is defined as:

$\begin{array}{cc}\prod _i\prod _j\exp \left(-\frac{{p_i-p_j}^2}{2{\sigma }_2^2}\right)\prod _i\exp \left(-\frac{{\sum _jp_j-4p_i}^2}{2{\sigma }_3^2}\right)& \left(3\right)\end{array}$

where σ₁,σ₂,σ₃ are the respective uncertainty measurements, and jεN(i) is the pixel locations of the first-order neighbors, that is surrounding pixels, of pixel i. The likelihood energy measures the faithfulness of the output dense normal map to the input O. The first exponent in the prior energy (3) enforces the smoothness of the normal orientation, while the second exponent minimizes the surface curvature corresponds to the output dense normal map. The values σ₁,σ₂,σ₃ can be set by the user, which indicates the weight or importance of the associated exponents. The energy function (1) is convex with the above likelihood and prior definitions, and the associated optimization thus converges to a global optimal solution. Standard numerical optimization packages can be used to solve equation (1). From this dense normal map, G, estimated using equation (1), a 3D surface or height map can be generated using standard shape-from-normal estimation techniques.

Illustrative Apparatus

FIG. 7 illustrates, at 700, one example of an apparatus 702 that can implement the discussed processes and techniques. While illustrated as a desktop computer, apparatus 702 can include any suitable device, including a personal computer, a laptop computer, a personal digital assistant (PDA), a cellular phone, and so on.

Apparatus 702 includes a system memory 704, processor(s) 706, and input/output components 708. Processor(s) 706 may include, for example, microprocessors, microcomputers, microcontrollers, multi-core processors, and so forth. Processor(s) 706 are configured to retrieve and execute computer-program instructions stored in system memory 704. System memory 704 includes computer-readable media in the form of volatile memory, such as Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash RAM. Input/output components 708 (e.g., a mouse and keyboard) provide data input and output capabilities for apparatus 702.

Embodied on system memory 704 are programming modules that can implement the discussed processes and techniques. Image canvas module 710 generates an image canvas and receives user markup of the image canvas. The image canvas module 710 can also import and export images as part of the image markup process. Shape palette module 712 generates and/or retrieves shape palettes and receives user markup of shape palettes. Rendering module 714 processes user markup of an image canvas and a shape palette to render an image with 3D characteristics. In some examples, rendering an image with 3D characteristics includes displaying the image. User interface module 716 generates and/or displays user interfaces that enable a user to interact with the apparatus and implement the discussed processes and techniques.

While various illustrative device and operating implementations have been described, the components, modules, and features of these implementations may be rearranged, modified, and/or may be omitted entirely, depending on the circumstances.

Also, it should be understood that certain acts in the methods need not be performed in the order described, may be rearranged, modified, and/or may be omitted entirely, depending on the circumstances.

Moreover, any of the acts described above with respect to any method may be implemented by a processor or other computing device based on instructions stored on one or more computer-readable media. Computer-readable media can be any available media that can be accessed locally or remotely by the resource modeling application. By way of example, and not limitation, computer-readable media may comprise volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data. Computer-readable media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by the resource modeling application. Combinations of the any of the above should also be included within the scope of computer-readable media.

Conclusion

Although the invention has been described in language specific to structural features and/or methodological acts, it is to be understood that the invention is not necessarily limited to the specific features or acts described. Rather, the specific features and acts are disclosed as illustrative forms of implementing the invention.

Claims

1. A method comprising: receiving a graphical image in a two-dimensional space;presenting a shape palette comprising a two-dimensional representation of a three-dimensional shape; andimparting characteristics of three-dimensionality to the graphical image based on user markup of the shape palette.

2. A method as recited in claim 1, wherein receiving the graphical image comprises loading an image file into the two-dimensional space.

3. A method as recited in claim 1, wherein the shape palette comprises an orthographic projection of a shape model.

4. A method as recited in claim 1, wherein the shape palette comprises a Lambertian-shaded image.

5. A method as recited in claim 1, wherein the shape palette comprises a sphere.

6. A method as recited in claim 1, wherein the shape palette comprises a plurality of pixels, each of the plurality of pixels associated with a normal value.

7. A method as recited in claim 1, wherein the user markup of the shape palette comprises user selection of a plurality of points on the shape palette, the plurality of points used to extract a set of pixels and a set of normal values from the shape palette, the imparting of the characteristics of three-dimensionality to the graphical image based on processing the set of pixels and the set of normal values to render a graphical image with three-dimensional characteristics.

8. A method as recited in claim 1, further comprising rendering the graphical image with the characteristics of three-dimensionality.

9. One or more computer-readable media comprising computer executable instructions that, when executed, perform acts comprising: generating a shape palette, the shape palette comprising a two-dimensional visual representation of geometric information in three dimensions;receiving user markup of the shape palette to extract a set of geometric information; andrendering a visual image based at least in part on the set of geometric information.

10. One or more computer-readable media as recited in claim 9, wherein the shape palette comprises an orthographic projection that incorporates Lambertian shading.

11. One or more computer-readable media as recited in claim 9, wherein the shape palette comprises one or more of a concave sphere and a convex sphere.

12. One or more computer-readable media as recited in claim 9, wherein the set of geometric information comprises a plurality of pixels and a plurality of normal values, each of the plurality of pixels associated with one of the plurality of normal values.

13. One or more computer-readable media as recited in claim 12, wherein rendering the visual image comprises deriving a dense normal map from the plurality of pixels and the plurality of normal values.

14. One or more computer-readable media as recited in claim 9, wherein rendering the visual image comprises: receiving user markup of a graphic image; andcorrelating the user markup of the graphic image with the user markup of the shape palette.

15. One or more computer-readable media as recited in claim 9, wherein the visual image is rendered based at least in part on a plurality of brightness values.

16. An apparatus comprising: a canvas module to generate a user canvas and to receive user markup on the user canvas; anda shape palette module to generate a shape palette and configured to receive user markup of the shape palette, the shape palette comprising a two-dimensional visual representation of three-dimensional geometric information, the user markup of the shape palette extracting a set of geometric information from the shape palette that is applied at least in part to render a graphical image.

17. An apparatus as recited in claim 16, wherein the shape palette comprises an orthographic projection incorporating Lambertian shading.

18. An apparatus as recited in claim 16, wherein the geometric information comprises a plurality of normal values.

19. An apparatus as recited in claim 16, further comprising a rendering module to process the set of geometric information to produce a data set and render an image with three-dimensional visual characteristics based at least in part on the data set.

20. An apparatus as recited in claim 16, further comprising a user interface module to display the user canvas and the shape palette.