Method and apparatus for supersampling based on the local rate of change in texture

Claim 2

A computer-implemented texture sampling method for supersampling multi-resolutional texture data to provide texture for an image to be displayed on a display screen in real-time, said multi-resolution texture data comprising a texture pattern and at least one reduced resolution texture pattern, said multi-resolution texture data being defined in each texture dimension of texture space by texture coordinates, and said image being represented by pixel data defined in screen space dimensions by screen space coordinates (x,y), said texture sampling method comprising the steps of: (a) determining a destination sample position in screen space; (b) determining texture coordinates in texture space corresponding to said destination sample position; (c) calculating local rates of change of said texture coordinates corresponding to said destination sample position; further comprising, prior to step (d), the step of scaling said local rate of change of at least one texture coordinate calculated in step (c), wherein each local rate of change for each texture coordinate is scaled independently; (d) determining first horizontal and vertical screen space sampling rates (sampX, sampY) based on said local rates of change in texture calculated in step (c) and independently scaled in said scaling step; (e) reducing said first horizontal and vertical screen space sampling rates (sampX, sampY) to second horizontal and vertical screen space sampling rates (newSampX, newSampY) corresponding to a reduced resolution texture image when said first horizontal and vertical screen space sampling rates are each greater than or equal to 2; and (f) sampling said texture data at a number of positions distributed according to said second horizontal and vertical screen space sampling rates to approximate a footprint corresponding to said destination sample position wherein said texture space comprises horizontal and vertical texture dimensions corresponding to horizontal and vertical texture coordinates (s,t), said texture pattern being provided as a video image, and wherein said scaling step scales local rates of change along said horizontal and vertical texture coordinates by different amounts to compensate for oversampling in said video image.

Claim 3

The method of claim 1, wherein said scaling step comprises multiplying said local rate of change of at least one texture coordinate by a corresponding, predetermined scaling factor approximately equal to twice the Nyquist frequency limit for said video image divided by an actual sampling rate of said video image.

Claim 4

A computer-implemented texture sampling method for supersampling multi-resolutional texture data to provide texture for an image to be displayed on a display screen in real-time, said multi-resolution texture data comprising a texture pattern and at least one reduced resolution texture pattern, said multi-resolution texture data being defined in each texture dimension of texture space by texture coordinates, and said image being represented by pixel data defined in screen space dimensions by screen space coordinates (x,y), said texture sampling method comprising the steps of: (a) determining a destination sample position in screen space; (b) determining texture coordinates in texture space corresponding to said destination sample position; (c) calculating local rates of change of said texture coordinates corresponding to said destination sample position; (d) determining first horizontal and vertical screen space sampling rates (sampX, sampY) based on said local rates of change in texture calculated in step (c); (e) reducing said first horizontal and vertical screen space sampling rates (sampX, sampY) to second horizontal and vertical screen space sampling rates (newSampX, newSampY) corresponding to a reduced resolution texture image when said first horizontal and vertical screen space sampling rates are each greater than or equal to 2; and (f) sampling said texture data at a number of positions distributed according to said second horizontal and vertical screen space sampling rates to approximate a footprint corresponding to said destination sample position, wherein said step (c) includes the step of: calculating a set of texture gradients (ds/dx, ds/dy, dt/dx, dt/dy) representing local rates of change in said texture coordinates (s,t) along horizontal and vertical screen space coordinates (x,y) through the footprint.

Claim 5

A computer-implemented texture sampling method for supersampling multi-resolutional texture data to provide texture for an image to be displayed on a display screen in real-time, said multi-resolution texture data comprising a texture pattern and at least one reduced resolution texture pattern, said multi-resolution texture data being defined in each texture dimension of texture space by texture coordinates, and said image being represented by pixel data defined in screen space dimensions by screen space coordinates (x,y), said texture sampling method comprising the steps of: (a) determining a destination sample position in screen space; (b) determining texture coordinates in texture space corresponding to said destination sample position; (c) calculating local rates of change of said texture coordinates corresponding to said destination sample position; (d) determining first horizontal and vertical screen space sampling rates (sampX, sampY) based on said local rates of change in texture calculated in step (c); (e) reducing said first horizontal and vertical screen space sampling rates (sampX, sampY) to second horizontal and vertical screen space sampling rates (newSampX, newSampY) corresponding to a reduced resolution texture image when said first horizontal and vertical screen space sampling rates are each greater than or equal to 2; and (f) sampling said texture data at a number of positions distributed according to said second horizontal and vertical screen space sampling rates to approximate a footprint corresponding to said destination sample position, wherein said step (c) includes the step of: calculating a set of texture gradients (ds/dx, ds/dy, dt/dx, dt/dy) representing local rates of change in said texture coordinates (s,t) along horizontal and vertical screen space coordinates (x,y) through the footprint.

Claim 6

The method of claim 3, wherein said step (d) comprises the steps of: determining a first horizontal screen space sampling rate (sampX) based on the maximum change across the pixel in s and t texture coordinates along the x screen space coordinate; and determining a first vertical screen space sampling rate (sampY) based on the maximum change across the pixel in s and t texture coordinates along the y screen space coordinate.

Claim 7

The method of claim 4, wherein sampX is defined as: sampX=power -- of -- 2 -- ceil(LENGTH(LODSX, LODTX), where LODSX equals a ds/dx texture gradient, LODTX equals a dt/dx texture gradient, LENGTH represents a distance function which approximates the square root of the quantity (LODSX) 2 +(LODTX) 2 !, and power-of-2-ceil represents a rounding function which rounds to the next nearest power of two; and sampY is defined as: sampY=power -- of -- 2 -- ceil(LENGTH(LODSY,LODTY), where LODSY equals a ds/dy texture gradient, LODTY equals a dt/dy texture gradient, LENGTH represents a distance function which approximates the square root of the quantity (LODSY) 2 +(LODTY) 2 !, and power-of-2-ceil represents a rounding function which rounds to the next nearest power of two.

Claim 8

The method of claim 5, wherein step (d) further comprises the steps of: calculating a footprint area A F based on a set of texture gradients ds/dx, dt/dy, ds/dy, and dt/dx where A F =ABS (ds/dx*dt/dy-ds/dy*dt/dx), ABS denotes an absolute value function; modifying A F when the pixel footprint area collapses to approximate a line, said modifying step consisting of setting A F to equal the maximum of ABS (ds/dx*dt/dy-ds/dy*dt/dx) and SampX+SampY; and calculating a final vertical screen space sampling rate (sampY) based on the ratio of A F , as determined after said modifying step, and SampX.

Claim 9

The method of claim 6, wherein said final vertical screen space sampling rate, SampY is defined as: sampY=power -- of -- 2 -- ceil (A F /SampX) where power -- of -- 2 -- ceil represents a rounding function which rounds to the next nearest power of two.

Claim 10

The method of claim 3, wherein said step (d) comprises the steps of: calculating a footprint area A F based on a set of texture gradients ds/dx, dt/dy, ds/dy, and dt/dx where A F =ABS (ds/dx*dt/dy-ds/dy*dt/dx), ABS denotes an absolute value function; determining a first horizontal screen space sampling rate (sampX) based on the maximum change across the pixel in s and t texture coordinates along the x screen space coordinate; and determining a first vertical screen space sampling rate (sampY) based on the ratio of A F and SampX.

Claim 11

The method of claim 8, wherein in said step (d): sampX is defined as: sampX=power -- of -- 2 -- ceil(LENGTH(LODSX, LODTX), where LODSX equals ds/dx, LODTX equals dt/dx, LENGTH represents a distance function which approximates the square root of the quantity (LODSX) 2 +(LODTX) 2 !, and power -- of -- 2 -- ceil represents a rounding function which rounds to the next nearest power of two; and sampY is defined as: sampY=power -- of -- 2 -- ceil (A F /SampX) where power -- of -- 2 -- ceil represents a rounding function which rounds to the next nearest power of two.

Claim 12

The method of claim 3, wherein said texture data comprises at least one of two-dimensional texture data and three-dimensional texture data.

Claim 13

A computer graphics rasterizer system for supersampling multi-resolutional texture data to provide texture for an image to be displayed on a display screen in real-time, said multi-resolution texture data comprising a texture pattern and at least one reduced resolution texture pattern, said multi-resolution texture data being defined in each texture dimension of texture space by texture coordinates, and said image being represented by pixel data defined in screen space dimensions by screen space coordinates (x,y), said rasterizer system comprising: destination sample position determining means for determining a destination sample position in screen space; texture coordinate determining means for determining texture coordinates in texture space corresponding to said destination sample position; local texture rate of change calculating means for calculating local rates of change of said texture coordinates corresponding to said destination sample position; screen space sampling rate determining means for determining first horizontal and vertical screen space sampling rates (sampX, sampY) based on said local rates of change in texture calculated by said local texture rate of change calculating means and reducing said first horizontal and vertical screen space sampling rates (sampX, sampY) to second horizontal and vertical screen space sampling rates (newSampX, newSampY) corresponding to a reduced resolution texture image when said first horizontal and vertical screen space sampling rates are each greater than or equal to 2; and sampling means for sampling said texture data at a number of positions distributed according to said second horizontal and vertical screen space sampling rates to approximate a footprint corresponding to said destination sample position; further comprising scaling means for scaling said local rate of change of at least one texture coordinate calculated in said local texture rate of change calculating means, wherein each local rate of change for each texture coordinate is scaled independently to allow independent control over said second horizontal and vertical screen space sampling rates; wherein said texture space comprises horizontal and vertical texture dimensions corresponding to horizontal and vertical texture coordinates (s,t), said texture pattern being provided as a video image, and wherein said scaling means comprises means for scaling said local rates of change along each texture coordinate by different amounts to compensate for oversampling in said video image.

Claim 14

A computer graphics rasterizer system for supersampling multi-resolutional texture data to provide texture for an image to be displayed on a display screen in real-time, said multi-resolution texture data comprising a texture pattern and at least one reduced resolution texture pattern, said multi-resolution texture data being defined in each texture dimension of texture space by texture coordinates, and said image being represented by pixel data defined in screen space dimensions by screen space coordinates (x,y), said rasterizer system comprising: destination sample position determining means for determining a destination sample position in screen space; texture coordinate determining means for determining texture coordinates in texture space corresponding to said destination sample position; local texture rate of change calculating means for calculating local rates of change of said texture coordinates corresponding to said destination sample position; screen space sampling rate determining means for determining first horizontal and vertical screen space sampling rates (sampX, sampY) based on said local rates of change in texture calculated by said local texture rate of change calculating means and reducing said first horizontal and vertical screen space sampling rates (sampX, sampY) to second horizontal and vertical screen space sampling rates (newSampX, newSampY) corresponding to a reduced resolution texture image when said first horizontal and vertical screen space sampling rates are each greater than or equal to 2; and sampling means for sampling said texture data at a number of positions distributed according to said second horizontal and vertical screen space sampling rates to approximate a footprint corresponding to said destination sample position; further comprising scaling means for scaling said local rate of change of at least one texture coordinate calculated in said local texture rate of change calculating means, wherein each local rate of change for each texture coordinate is scaled independently to allow independent control over said second horizontal and vertical screen space sampling rates; wherein said texture space comprises horizontal and vertical texture dimensions corresponding to horizontal and vertical texture coordinates (s,t), said texture pattern being provided as a video image, and wherein said scaling means comprises means for scaling said local rates of change along each texture coordinate by different amounts to compensate for oversampling in said video image.

Claim 15

The rasterizer system of claim 11, wherein said scaling means further comprises multiplying means for multiplying said local rate of change of at least one texture coordinate by a corresponding, predetermined scaling factor approximately equal to twice the Nyquist frequency limit for said video image divided by an actual sampling rate of said video image.

Claim 16

A computer graphics rasterizer system for supersampling multi-resolutional texture data to provide texture for an image to be displayed on a display screen in real-time, said multi-resolution texture data comprising a texture pattern and at least one reduced resolution texture pattern said multi-resolution texture data being defined in each texture dimension of texture space by texture coordinates, and said image being represented by pixel data defined in screen space dimensions by screen space coordinates (x,y), said rasterizer system comprising: destination sample position determining means for determining a destination sample position in screen space; texture coordinate determining means for determining texture coordinates in texture space corresponding to said destination sample position; local texture rate of change calculating means for calculating local rates of change of said texture coordinates corresponding to said destination sample position; screen space sampling rate determining means for determining first horizontal and vertical screen space sampling rates (sampX, sampY) based on said local rates of change in texture calculated by said local texture rate of change calculating means and reducing said first horizontal and vertical screen space sampling rates (sampX, sampY) to second horizontal and vertical screen space sampling rates (newSampX, newSampY) corresponding to a reduced resolution texture image when said first horizontal and vertical screen space sampling rates are each greater than or equal to 2; and sampling means for sampling said texture data at a number of positions distributed according to said second horizontal and vertical screen space sampling rates to approximate a footprint corresponding to said destination sample position, wherein said local texture rate of change calculating means comprises texture gradient set calculating means for calculating a set of texture gradients (ds/dx, ds/dy, dt/dx,dt/dy) representing local rates of change in said texture coordinates (s,t) along horizontal and vertical screen space coordinates (x,y) through the footprint.

Claim 17

A computer graphics rasterizer system for supersampling multi-resolutional texture data to provide texture for an image to be displayed on a display screen in real-time, said multi-resolution texture data comprising a texture pattern and at least one reduced resolution texture pattern said multi-resolution texture data being defined in each texture dimension of texture space by texture coordinates, and said image being represented by pixel data defined in screen space dimensions by screen space coordinates (x,y), said rasterizer system comprising: destination sample position determining means for determining a destination sample position in screen space; texture coordinate determining means for determining texture coordinates in texture space corresponding to said destination sample position; local texture rate of change calculating means for calculating local rates of change of said texture coordinates corresponding to said destination sample position; screen space sampling rate determining means for determining first horizontal and vertical screen space sampling rates (sampX, sampY) based on said local rates of change in texture calculated by said local texture rate of change calculating means and reducing said first horizontal and vertical screen space sampling rates (sampX, sampY) to second horizontal and vertical screen space sampling rates (newSampX, newSampY) corresponding to a reduced resolution texture image when said first horizontal and vertical screen space sampling rates are each greater than or equal to 2; and sampling means for sampling said texture data at a number of positions distributed according to said second horizontal and vertical screen space sampling rates to approximate a footprint corresponding to said destination sample position, wherein said local texture rate of change calculating means comprises texture gradient set calculating means for calculating a set of texture gradients (ds/dx, ds/dy, dt/dx,dt/dy) representing local rates of change in said texture coordinates (s,t) along horizontal and vertical screen space coordinates (x,y) through the footprint.

Claim 18

The rasterizer system of claim 13, wherein said screen space sampling rate determining means comprises: means for determining a first horizontal screen space sampling rate (sampX) based on the maximum change across the pixel in s and t texture coordinates along the x screen space coordinate; and means for determining a first vertical screen space sampling rate (sampY) based on the maximum change across the pixel in s and t texture coordinates along the y screen space coordinate.

Claim 19

The rasterizer system of claim 14, wherein sampX is defined as: sampX=power -- of -- 2 -- ceil(LENGTH(LODSX,LODTX), where LODSX equals a ds/dx texture gradient, LODTX equals a dt/dx texture gradient, LENGTH represents a distance function which approximates the square root of the quantity (LODSX) 2 +(LODTX) 2 !, and power-of-2-ceil represents a rounding function which rounds to the next nearest power of two; and sampY is defined as: sampY=power -- of -- 2 -- ceil(LENGTH(LODSY,LODTY), where LODSY equals a ds/dy texture gradient, LODTY equals a dt/dy texture gradient, LENGTH represents a distance function which approximates the square root of the quantity (LODSY) 2 +(LODTY) 2 !, and power-of-2-ceil represents a rounding function which rounds to the next nearest power of two.

Claim 20

The rasterizer system of claim 14, wherein said screen space sampling rate determining means comprises: means for calculating a footprint area A F based on a set of texture gradients ds/dx, dt/dy, ds/dy, and dt/dx where A F =ABS (ds/dx*dt/dy-ds/dy*dt/dx), ABS denotes an absolute value function; means for modifying A F when the pixel footprint area collapses to approximate a line, said modifying step consisting of setting A F to equal the maximum of ABS (ds/dx*dt/dy-ds/dy*dt/dx) and SampX+SampY; and means for calculating a final vertical screen space sampling rate (sampY) based on the ratio of A F , as determined by said modifying means, and SampX.

Claim 21

The rasterizer system of claim 16, wherein said final vertical screen space sampling rate, SampY, is defined as: sampY=power -- of -- 2 -- ceil (A F /SampX) where power -- of -- 2 -- ceil represents a rounding function which rounds to the next nearest power of two.

Claim 22

The rasterizer system of claim 13, wherein said screen space sampling rate determining means comprises: means for calculating a footprint area A F based on a set of texture gradients ds/dx, dt/dy, ds/dy, and dt/dx where A F =ABS (ds/dx*dt/dy-ds/dy*dt/dx), ABS denotes an absolute value function; means for determining a first horizontal screen space sampling rate (sampX) based on the maximum change across the pixel in s and t texture coordinates along the x screen space coordinate; and means for determining a first vertical screen space sampling rate (sampY) based on the ratio of A F and SampX.

Claim 23

The rasterizer system of claim 18, wherein sampX is defined as: sampX=power -- of -- 2 -- ceil(LENGTH(LODSX, LODTX), where LODSX equals ds/dx, LODTX equals dt/dx, LENGTH represents a distance function which approximates the square root of the quantity (LODSX) 2 +(LODTX) 2 !, and power -- of -- 2 -- ceil represents a rounding function which rounds to the next nearest power of two; and sampY is defined as: sampY=power -- of -- 2 -- ceil (A F /SampX) where power -- of -- 2 -- ceil represents a rounding function which rounds to the next nearest power of two.

Claim 24

The rasterizer system of claim 13, wherein said texture data comprises at least one of two-dimensional texture data and three-dimensional texture data.

Claim 25

The rasterizer system of claim 13, wherein said multi-resolutional data comprises an image pyramid of reduced resolution images progressively filtered by a power of two.

Claim 26

A computer-implemented texture processing method for mapping texture data corresponding to a texture pattern comprising the steps of: (a) determining a destination sample position in screen space for a pixel; (b) determining texture coordinates in texture space corresponding to said destination sample position; (c) calculating local rates of change of said texture coordinates corresponding to said destination sample position; (d) independently scaling said local rates of change for each texture coordinate calculated in step (c); and (e) calculating a level-of-detail for texture data to be mapped from a pixel footprint in texture space to cover a pixel at said destination sample position in screen space pixel, said level-of-detail being calculated based on said independently scaled local rates of change in texture coordinates; whereby, an anisotropic texture level-of-detail compensation is provided which allows independent control of the effect of each texture coordinate on the level-of-detail of texture data chosen to be mapped to a pixel; wherein said texture space comprises horizontal and vertical texture dimensions corresponding to horizontal and vertical texture coordinates (s,t), said texture pattern being provided as a video image, and wherein said scaling step (d) scales local rates of change along each texture coordinate by different amounts to compensate for oversampling in said video image.

Claim 27

A computer-implemented texture processing method for mapping texture data corresponding to a texture pattern comprising the steps of: (a) determining a destination sample position in screen space for a pixel; (b) determining texture coordinates in texture space corresponding to said destination sample position; (c) calculating local rates of change of said texture coordinates corresponding to said destination sample position; (d) independently scaling said local rates of change for each texture coordinate calculated in step (c); and (e) calculating a level-of-detail for texture data to be mapped from a pixel footprint in texture space to cover a pixel at said destination sample position in screen space pixel, said level-of-detail being calculated based on said independently scaled local rates of change in texture coordinates; whereby, an anisotropic texture level-of-detail compensation is provided which allows independent control of the effect of each texture coordinate on the level-of-detail of texture data chosen to be mapped to a pixel; wherein said texture space comprises horizontal and vertical texture dimensions corresponding to horizontal and vertical texture coordinates (s,t), said texture pattern being provided as a video image, and wherein said scaling step (d) scales local rates of change along each texture coordinate by different amounts to compensate for oversampling in said video image.

Claim 28

The method of claim 22, wherein said scaling step (d) further comprises multiplying said local rate of change of at least one texture coordinate by a corresponding, predetermined scaling factor approximately equal to twice the Nyquist frequency limit for said video image divided by an actual sampling rate of said video image.