System, method and article of manufacture for pixel shaders for programmable shading

[0002] This application is related to a co-pending application entitled “GRAPHICS PIPELINE INCLUDING COMBINER STAGES” filed Mar. 22, 1999 naming David B. Kirk, Matthew Papakipos, Shaun Ho, Walter Donovan, and Curtis Priem as inventors, and issued as U.S. Pat. No. 6,333,744, and which is incorporated herein by reference in its entirety.

[0003] 1. Field of the Invention

[0004] The present invention relates to computer graphics, and more particularly to texture sampling in a computer graphics processing pipeline.

[0005] 2. Background of the Invention

[0006] Recent advances in computer performance have enabled graphic systems to provide more realistic graphical images using personal computers and home video game computers. In such graphic systems, some procedure must be implemented to “render” or draw graphic primitives to the screen of the system. A “graphic primitive” is a basic component of a graphic picture, such as a polygon, e.g., a triangle, or a vector. All graphic pictures are formed with combinations of these graphic primitives. Many procedures may be utilized to perform graphic primitive rendering.

[0007] Early graphic systems displayed images representing objects having extremely smooth surfaces. That is, textures, bumps, scratches, or other surface features were not modeled. In order to improve the quality of the image, texture mapping was developed to model the complexity of real world surface images. In general, texture mapping is the mapping of an image or a function onto a surface in three dimensions. Texture mapping is a relatively efficient technique for creating the appearance of a complex image without the tedium and the high computational cost of rendering the actual three dimensional detail that might be found on a surface of an object.

[0008] Prior Art FIG. 1 illustrates a graphics pipeline with which texture mapping may be performed. As shown, included is a transform engine 100, a set-up module 102, a rasterizer 104, a texture math module 106, a level of detail (LOD) calculator 108, a texture fetch module 110, a texture filter 112, and a texture combination engine 114. It should be noted that the transform engine 100 and set-up module 102 need not necessarily be required in the graphics pipeline of a graphics integrated circuit.

[0009] During operation, the transform engine 100 may be used to perform scaling, rotation, and projection of a set of three dimensional vertices from their local or model coordinates to the two dimensional window that will be used to display the rendered object. The setup module 102 utilizes the world space coordinates provided for each triangle to determine the two dimensional coordinates at which those vertices are to appear on the two dimensional window. Prior Art FIG. 2 illustrates the coordinates 200 of the vertices 201 which define a triangle 202. If the vertices 201 of the triangle 202 are known in screen space, the pixel positions vary along scan lines within the triangle 202 in screen space and may be determined.

[0010] The setup module 102 and the rasterizer module 104 together use the three dimensional world coordinates to determine the position of each pixel contained inside each of the triangles. Prior Art FIG. 3 illustrates a plurality of pixels 300 identified within the triangle 202 in such a manner. The color values of the pixels in the triangle 202 vary from pixel to pixel in world space. During use, the setup module 102 and the rasterizer module 104 generate interpolated colors, depth and texture coordinates.

[0011] The setup module 102 and the rasterizer module 104 then feed the pixel texture coordinates to the texture math module 106 to determine the appropriate texture map colors. In particular, texture coordinates are generated that reference a texture map using texture coordinate interpolation which is commonly known to those of ordinary skill in the art. This is done for each of the pixels 300 identified in the triangle 202. Prior Art FIG. 3 illustrates texture coordinates 302 for the pixels 300 identified within the triangle 202.

[0012] Next, a LOD calculation is performed by the LOD calculator 108. Occasionally during rendering, one texel, or texture element, will correspond directly to a single pixel that is displayed on a monitor. In this situation the level of detail (LOD) is defined to be equal to zero (0) and the texel is neither magnified nor minified. However, the displayed image can be a magnified or minified representation of the texture map. If the texture map is magnified, multiple pixels will represent a single texel. A magnified texture map corresponds to a negative LOD value. If the texture map is minified, a single pixel represents multiple texels. A minified texture map corresponds to a positive LOD value. In general, the LOD value corresponds to the number of texels in the texture map ‘covered’ by a single pixel.

[0013] The amount of detail stored in different LOD representations may be appreciated by drawing an analogy to the detail perceived by an observer while observing a texture map. For example, very little detail may be perceived by an observer while watching an automobile from a distance. On the other hand, several details such as doors, windows, mirrors will be perceived if the observer is sufficiently close to the automobile. A finer level LOD will include such additional details, and a courser LOD will not.

[0014] Once the appropriate level of detail of the texture map is selected based on the calculated LOD value, the texture coordinates generated by the texture math module 106 are used to fetch the appropriate texture map colors using the texture fetch module 110. These texture map colors are then filtered by the texture filter module 112. The combiner engine 114 combines together the various colors and textures fetched by the texture fetch module 110 and filtered by the texture filter module 112.

[0015] It is important to note that the pipeline described hereinabove has a linear topology. These and other simplistic non-linear pipelines only enable one texture fetch and texture calculation per rendering pass. This is a limited design that is static in nature. There is thus a need for a pipeline that allows for more dynamic texture fetches and shading calculations, and in particular, the ability for feeding filter results back to influence subsequent texture address calculations.

[0016] A system, method and article of manufacture are provided for interweaving shading calculations and texture retrieval operations during texture sampling in a graphics pipeline. First, a shading calculation is performed in order to generate output, i.e. colors or texture coordinates. Next, texture information is retrieved, and another shading calculation is performed using the texture information in order to generate additional output. Texture information may be retrieved and shading calculations may then be repeated as desired. Thereafter, the generated output may be combined. As such, the repeated texture information retrieval and shading calculations may be carried out in an iterative, programmable manner.

[0017] In one embodiment of the present invention, edge distances of a primitive may be calculated, and at least one of the shading calculations involves the edge distances. Further, the shading calculation may include the calculation of a plurality of weighted coefficients from the edge distances. As an option, such weighted coefficients may include barycentric weights which use parameter values of the primitive to perform parameter interpolation.

[0018] In another embodiment of the present invention, the texture information may include filtered texture color information. As an option, the filtered texture value may be used as texture coordinates for use in retrieving further texture information when the texture information retrieval is repeated. Further, the repeated shading calculation may also use the output in order to generate additional output.

[0019] In still another embodiment of the present invention, the output may include diffuse output colors, fog output values, specular output colors, depth output values, texture color output values, a level of detail (LOD) value, and/or a Z-slope value. As an option, the shading calculation may include the calculation of a level of detail (LOD) which may occur after the texture information is retrieved.

[0020] In order to accomplish the foregoing, a graphics pipeline system may be provided which includes a shading module for performing the shading calculation in order to generate output. Coupled to the shading module is a texture look-up module for retrieving texture information. Further, a feedback loop is coupled between an input and an output of the shading module for performing additional shading calculations using the texture information from the texture look-up module. Also included is a combiner module coupled to the output of the shading module for combining the output generated by the shading module. In one aspect of the present embodiment, at least a pair of texture look-up modules is coupled to a pair of shading modules which together constitute at least four logical modules.

[0021] These and other advantages of the present invention will become apparent upon reading the following detailed description and studying the various figures of the drawings.

[0022] The foregoing and other aspects and advantages are better understood from the following detailed description of a preferred embodiment of the invention with reference to the drawings, in which:

[0023] Prior Art FIG. 1 illustrates a graphics pipeline with which texture mapping may be performed;

[0024] Prior Art FIG. 2 illustrates the coordinates of the vertices which define a triangle;

[0025] Prior Art FIG. 3 illustrates a plurality of pixels identified within the triangle of FIG. 2;

[0026] FIG. 4 illustrates a hardware implementation for programmable shading in accordance with one embodiment of the present invention;

[0027] FIG. 5 illustrates a graphical representation of the edge distances generated by the rasterizer in accordance with one embodiment of the present invention;

[0028] FIG. 6 illustrates a flow diagram depicting the manner in which the shading calculations are interweaved with the texture fetch operations in a plurality of iterations in accordance with one embodiment of the present invention;

[0029] FIG. 6A graphically illustrates the derivative values, (u_x) and (u_y), with which a level of detail calculation may be performed in accordance with one embodiment of the present invention; and

[0030] FIG. 7 illustrates a point within a primitive that is capable of being defined by barycentric weights.

[0031] FIGS. 1-3 illustrate the prior art. FIG. 4 shows an illustrative hardware implementation of the present invention. As shown, included is a set-up module 402, a rasterizer 404, a shader module 406, a texture fetch module 408, and a combiner 410 coupled to form a portion of a graphics processing pipeline 400. For reasons that will soon become apparent, a feedback loop 409 is coupled between an output of the shader module 406 to an input thereof. It should be noted that the set-up module 402, rasterizer 404, and combiner 410 operate in a conventional manner as set forth during reference to FIG. 1. While the combiner 410 may be implemented in any desired manner, one exemplary implementation is disclosed in a co-pending application entitled “GRAPHICS PIPELINE INCLUDING COMBINER STAGES” filed Mar. 22, 1999 naming David B. Kirk, Matthew Papakipos, Shaun Ho, Walter Donovan, and Curtis Priem as inventors, and issued as U.S. Pat. No.: 6,333,744, and which is incorporated herein by reference in its entirety.

[0032] With continuing reference to FIG. 4, the various inputs and outputs are shown for each of the components. As shown, the rasterizer 404 generates edge distances which are used by the shader module 406 and texture fetch module 408.

[0033] Also shown in FIG. 4 is an optional feedback first-in first-out (FIFO) buffer. When the feedback loop 409 is not utilized, the temporary data calculated internally by the present invention may be dropped before being sent to the texture fetch module 408.

[0034] As an option, however, the shader module 406 may be reused, and some of this data (like the barycentric coordinates) may be reused each time a particular group of pixels, or “quad,” goes through the shader module 406. If new colors are generated during one pass, these colors may continuously be associated with the quad on subsequent passes. Further, more than one triangle may be processed at a time while employing the feedback loop 409, since data from several triangles generally appears while waiting for the texture fetch module 408 to calculate an individual texture look-up.

[0035] To address this, the loopback FIFO 407 may be utilized to hold barycentric weights, colors from previous passes, triangle information, and additional scheduling information to keep track of what the shader module 406 is supposed to do each pass. The FIFO 407 may include a plurality of bits that can be reconfigured to store whatever piece of data is appropriate. When the texture requests for a particular quad are sent to the texture fetch module 408, the associated data may also be placed in the FIFO 407. When the texture requests complete, the results may be combined with the data from the FIFO 407, and a small portion of logic may decide whether to send the completed quad to the combiner 410, or back around for another pass at the shader module 406.

[0036] FIG. 5 illustrates a graphical representation of the edge distances generated by the rasterizer 404. As shown, the edge distances (e₀, e₁, e₂) represent a perpendicular distance 500 from an edge 502 of a primitive 504 to a pixel 506 within the primitive 504. Such edge distances (e₀, e₁, e₂) thus identify the location of the pixel 506 within the primitive 504.

[0037] As a function of the shading calculations, various texture look-up operations may be carried out utilizing the texture look-up module 408 in order to obtain output having appropriate texture map colors. To accomplish this, texture coordinates may be sent to the texture look-up module 408. In response thereto, texture information is received from the texture look-up module 408. Such texture information may take any form including, but not limited to filtered texture color, etc.

[0038] During the course of use, the feedback loop 409 may be used for performing another shading calculation using the texture information from the texture look-up module 408 in order to generate further output. As an option, the texture information may include filtered texture look-up values for use in retrieving further texture information when the texture information retrieval operation is repeated. The present invention thus allows repeated texture information retrieval and shading calculations in an iterative, programmable-manner. In other words, each iteration may be programmed to do a desired shading calculation with or without a texture look-up, where each subsequent iteration may use results of previous texture look-ups for generating further results.

[0039] In one embodiment of the present invention, at least a pair of texture look-up modules is coupled to a pair of shading modules which together constitute at least four logical modules. Further, the system may be capable of performing both interpolation and shading calculations including pre-texture shading calculations and post-texture shading calculations

[0040] FIG. 6 illustrates a flow diagram depicting the manner in which the shading calculations are interweaved with the texture fetch operations in a plurality of iterations 600. As shown, each iteration 600 includes a shading calculation 602. During each shading calculation 602, a decision 604 may be made as to whether a texture look-up may be performed. If so, texture information may be retrieved during a texture look-up operation 605. Also during each iteration 600, another decision 606 is made as to whether an additional shading operation 602 is to be executed.

[0041] If it is determined in decision 606 that an additional shading operation 602 is to be performed, another iteration 600 is executed. On the other hand, if no further shading operations are to be executed, the process of FIG. 6 may be terminated. During use, the number of iterations may vary per the desires of the user.

[0042] As such, decision 604 allows additional texture information to be retrieved in subsequent shading calculations 600 based on previously retrieved texture information. It should be also noted that this may be done on different texture maps. In the alternative, it may be decided in decision 604 to not do a texture look-up, and merely perform a shading calculation 600 during a current iteration.

[0043] As mentioned earlier during reference to FIG. 4, the shading module 406 may carry out many various types.of operations in order to produce output of various types based on the edge distances (e₀, e₁, e₂) generated by the rasterizer 404. Such output may include, but are not limited to diffuse output colors, fog output values, specular output colors, depth output values, texture color output values, a level of detail (LOD) value, and/or a Z-slope value. As an option, the calculation of a level of detail (LOD) may be performed based on the texture information that is previously retrieved.

[0044] In one embodiment, the LOD is calculated numerically. For example, the LOD may be calculated in the manner set forth in a co-pending application entitled “SYSTEM, METHOD AND ARTICLE OF MANUFACTURE FOR CALCULATING A LEVEL OF DETAIL (LOD) DURING COMPUTER GRAPHICS PROCESSING” filed Mar. 24, 2000 under application Ser. No. 09/535,045, naming Walt Donovan as inventor which is incorporated herein by reference in its entirety.

[0045] In particular, four texture coordinates (u₀, u₁, u₂, u₃) may be first identified on the texture map corresponding to four portions of a block of pixels. Such texture coordinates may be such that u₀is an upper left corner of the block, u₁is an upper right corner of the block, u₂is a lower left corner of the block, and u₃is a lower right corner of the block. In one embodiment, the block may include a 2×2 block of pixels.

[0046] Thereafter, a derivative value (u_x), or du/dx, and (u_y), or du/dy, are calculated. FIG. 6A graphically illustrates the derivatives values (u_x) and (u_y). The pixel 607 is mapped in a texture coordinate space 608 with the derivatives values (u_x) and (u_y) defined as shown.

[0047] While the derivative values may be calculated in any known manner, the derivative value (u_x), or du/dx, and (u_y), or dv/dx, may be estimated using Equations #1.

[0048] u_x=((u₁−u₀)+(u₃−u₂))/2

[0049] u_y=((u₂−u₀)+(u₃−u₁))/2

[0050] It should be noted that similar calculations may be made for v_x,yand p_x,ytexture coordinates using the corresponding equations for derivatives with respect to the x and y axis. With such information, a LOD value is calculated using Equations #2.

[0051] LOD=½log₂(max (d_x, d_y)), where

[0052] d_x=u_x²+v_x²+p_x²

[0053] d_y=u_y²+v_y²+p_y²

[0054] In order to accomplish the foregoing shading operations set forth in FIGS. 4 and 6, perspective corrected barycentric weights (g₀, g₁, g₂) may be calculated from the edge distances (e₀, e₁, e₂). In a first embodiment, Equations #3 are utilized to calculate the perspective barycentric weights (g₀, g₁, g₂).

[0055] g0=e0*d

[0056] g1=e1*d

[0057] g2=e2*d,

[0058] where

[0059] s=e0*w0*w1+e1*w1*w2+e2*w2*w0

[0060] d=1/s,

[0061] where

[0062] w0, w1 and w2 are the perspective correction factors used to perform hyperbolic or perspective corrected interpolation

[0063] FIG. 7 illustrates a point within a primitive that is capable of being defined by barycentric weights (g₀, g₁, g₂). In particular, point (p) may be defined by Equation #4 in terms of the barycentric weights (g₀, g₁, g₂) and vertices (a, b, c) of the primitive. As an option, the perspective barycentric weights may use parameter values of the primitive in order to perform perspective corrected interpolation of vertex parameter values. Optionally, the perspective barycentric weights may use unclipped parameter values.

[0064] p=a*g₀+b*g₁+c*g₂

[0065] Given the perspective corrected barycentric weights (g₀, g₁, g₂), the various shading calculations may be performed using the equations set forth in Table 1.

[0066] Z in Screen Space(depth)—Calculate screen space z values at the vertices, then interpolate them per-pixel using the edge values as non-perspective corrected weights.

[0067] zs0=zc0/wc0; zs1=zc1/wc1; zs2=zc2/wc2.

[0068] zs=(e0zs0+e1zs1+e2zs2)/(e0+e1+e2)

[0069] Fog—Interpolate the fog range (may be affine xform of z eye, or distance from eye point—both computed per vertex).

[0070] Call fog range from xform fr.

[0071] fr=g0*fr0+g1*fr1+g2*fr2

[0072] then retrieve frac(fr) and run it through either:

[0073] 1) no table

[0074] 2) exp table

[0075] 3) exp{circumflex over ( )}2 table

[0076] Note: This table could be implemented as a texture map lookup. This would allow one to do an Openg1 fog table.

[0077] 4 (or any other number of) Non-projective 2-D Texture Coordinates—This optimization can only be done if all q's are one. Otherwise, the 2 projective case below is performed.

[0078] s=g0*s0+g1*s1+g2*s2

[0079] t=g0*t0+g1*t1+g2*t2

[0080] 2-D

[0081] sq=g0*s0+g1*s1+g2*s2

[0082] tq=g0*t0+g1*t1+g2*t2

[0083] q=g0*q0+g1*q1+g2*q2,

[0084] where

[0085] qi=1/q

[0086] s=sq*qi

[0087] t=tq*qi

[0088] Cube

[0089] sr=g0*s0+g1*s1+g2*s2

[0090] tr=g0*t0+g1*t1+g2*t2

[0091] rr=g0*r0+g1*r1+g2*r2

[0092] f=pickmax(s,t,r)

[0093] Note: f is a face index beginning at zero (for s). Pick is a function that chooses the fth entry from the list passed to it, where f is first parameter.

[0094] sr=pick(f,tr,rr,sr)

[0095] tr=pick(f,rr,sr,tr)

[0096] r=pick(f,sr,tr,rr),

[0097] where

[0098] ri=1/r

[0099] s=sr*ri

[0100] t=tr*ri

[0101] sq=g0*s0+g1*s1+g2*s2

[0102] tq=g0*t0+g1*t1+g2*t2

[0103] rq=g0*r0+g1*r1+g2*r2

[0104] q=g0*q0+g1*q1+g2*q2,

[0105] where

[0106] qi=1/q

[0107] s=sq*qi

[0108] t=tq*qi

[0109] r=rq*qi

[0110] 2-D dependent texture lookup—After the first texture lookup, two components of the argb color are reinterpreted as texture coordinates, and looked up again.

[0111] Dx6 Bump Mapping—After the first texture look-up, color r0,g0,b0 is received which is multiplied by 2×2 basis matrix, which is constant. s1 and t1 are the interpolated texcoords for the second look-up.

[0112] s1p=m11*r0+m12*g0+s1

[0113] t1p=m21*r0+m22*g0+t1

[0114] After the second texture lookup, received is r1,g1,b1,a1.

[0115] f=(b0*m31+m32)

[0116] r1p=r1*f

[0117] g1p=b1*f

[0118] b1p=b1*f

[0119] a1p=a1

[0120] Polygon Offset—let the notation z(1,0) indicate the z value of the pixel in the bottom right corner of the pixel block, or quad. z(0,1) would be the top left. compute the z slopes:

[0121] zx0=z(1,0)−z(0,0)

[0122] zy0=z(0,1)−z(0,0)

[0123] factor=max(abs(zx0),abs(zy0))

[0124] compute a new z:

[0125] zo=z+factor*zs+units, where factor and units are state. Loaded with pipelined state bundles.

[0126] Dot Product-based texture mapping—Using s0, t0, a first texture look-up is performed. Call the results a0 b0 c0. Dot products are taken between these values and the subsequent texture coordinates to generate a new set of texture coordinates for a subsequent texture lookup:

[0127] sp=s1*a0+t1*b0+r1*c0

[0128] tp=s2*a0+t2*b0+r2*c0

[0129] 2-D texture lookup performed using (sp, tp).

[0130] or

[0131] sp=s1*a0+t1*b0+r1*c0

[0132] tp=s2*a0+t2*b0+r2*c0

[0133] rp=s3*a0+t3*b0+r3*c0

[0134] 3-D texture lookup performed using (sp, tp, rp) or

[0135] Cube Texture coordinates performed (as above) using (sp, tp, rp)

[0136] Reflective Bump Mapping—Using s0, t0, a first texture look-up is performed. Call the results hs,ht,hr. this is the normal in tangent space.

[0137] interpolate s1, t1, r1.—this is the surface tangent vector in eye space

[0138] interpolate s2, t2, r2—this is the surface binormal vector.

[0139] interpolate s3, t3, r3—this is the surface normal vector.

[0140] These are used as a basis matrix by which to multiply the vector hs,ht,hr. This will give the normal in eye space.

[0141] so,

[0142] nx=s1*hs+s2*ht+s3*hr;

[0143] ny=t1*hs+t2*ht+t3*hr;

[0144] nz=r1*hs+r2*ht+r3*hr;

[0145] Use the (nx,ny,nz) vector as a cube map lookup for the diffuse lighting component.

[0146] Now compute the reflection vector per pixel.

[0147] let ne=nx*ex+ny*ey+nz*ez;

[0148] let n2=nx*nx+ny*ny+nz*nz;

[0149] rx=2*nx*ne/n2−ex;

[0150] ry=2*ny*ne/n2−ey;

[0151] rz=2*nz*ne/n2−ez;

[0152] Use this reflection vector as a cube map lookup.

[0153] Depth Texture Mapping with Dot Products—Using s0, t0, a first texture look-up is performed. Call the results a0, b0, c0. Dot products are performed between these values and two subsequent sets of texture coordinates to produce z clip and w clip values. These quotient of these values replaces the previously calculated z screen value.

[0154] zc=a0*s1+b0*t1+c0*r1;

[0155] wc=a0*s2+b0*t2+c0*r2;

[0156] zs=zc/wc

[0157] Pixel culling—The s, t, r, and q coordinates for a particular texture are interpolated per-pixel. Each coordinate is individually configured to check for either negative or non-negative values. If the texture coordinate matches the configuration, the pixel is culled (not drawn).

[0158] Isotropic BRDF—The results of two texture coordinate lookups are interpreted as 16-bit numbers, h0, 10 and h1, 11. A third texture lookup is performed using (h0, h1, 10−11).

[0159] It should be understood that each of the options set forth in the foregoing tables may be adapted to reuse common portions of the hardware set forth in FIG. 4. As set forth earlier, such hardware is capable of interpolating and performing texture address calculations during general operation.

[0160] Table 1 is based on perspective corrected barycentric weights (g₀, g₁, g₂) In another embodiment, non-perspective corrected barycentric weights (g₀, g₁, g₂) may be utilized which are defined in Equations #5. Non-perspective corrected barycentric weights replace perspective correct weights when texture coordinates, colors, depth, or fog are being interpolated.

[0161] g0=e0*d

[0162] g1=e1*d

[0163] g2=e2*d,

[0164] where

[0165] s=e0+e1+e2

[0166] d=1/s

[0167] While various embodiments have been described above, it should be understood that they have been presented by way of example only, and not limitation. Thus, the breadth and scope of a preferred embodiment should not be limited by any of the above described exemplary embodiments, but should be defined only in accordance with the following claims and their equivalents.