CN106023300A - Body rendering method and system of semitransparent material - Google Patents

Abstract

The invention discloses a body rendering method of a semitransparent material. The method comprises the following steps: step one, obtaining light source attribute data and material parameter data of a semitransparent module body; step two, for each light color spectrum, through combination with the light source attribute data and the material parameter data of the semitransparent module body, generating a body light quantity model; and step three, based on a conventional light projection algorithm, an influence exerted by voxels around a sampling point on the sampling point is added, and a rendering formula of the semitransparent module body is obtained. The invention also discloses a body rendering system of a semitransparent material, comprising structures of such three portions, i.e., a data collection unit, a body light quantity model generation unit and a rendering unit. According to the invention, little extra memory is consumed based on a conventional light projection three-dimensional rendering algorithm, real-time subsurface scattering simulation is realized, and the problem of too much consumption of time and memory in conventional simulation subsurface scattering is well solved.

Description

The body rendering intent of a kind of translucent material and system

Technical field

The present invention relates to graphics rendering technology field, particularly relate to one and facilitate the translucent material Subsurface Scattering side of rendering Method and system.

Background technology

Along with the development of three-dimensional visualization technique, people are more and more stronger to the cognitive sense of virtual reality.Thing in actual environment The lighting effect that body presents can be different because of the difference of object material, must take into scene when therefore rendering virtual scene true to nature In the material properties of each object.

In numerous materials, there is the true to nature of translucent attribute material and render complexity the most, most challenge.Wash with watercolours in high quality The not only requirement of dye translucent material model considers the absorption between light and three-dimensional voxel and reflection effect, it is necessary to consider light Scattering effect between voxel, this under model surface produce scattering effect we be referred to as " Subsurface Scattering ", refer in particular to Light enters object from body surface point, through scattering-in, finally from the light transmittance process of other outgoing of body surface.Secondary Surface scattering shows particularly evident in translucent material, such as human skin, wax candle, the material such as jade and snow, visual effect Be mainly reflected in the feature of softening object, interior of articles occur color bleeding (color bleeding) and shadow edge have by The phenomenon of gradual transition.Such as, human ear shade owing to Subsurface Scattering can present micro-red color rather than completely black shade.

Owing to Subsurface Scattering becomes extremely complex by light path, traditional three-dimensional visualized algorithm ray-tracing is not The rendering effect that high-quality is true to nature can be generated, though and Monte Carlo ray tracing algorithm can render Subsurface Scattering light well According to effect, but time complexity is high, renders a width virtual scene figure and generally requires the time of a couple of days, and cost is the biggest.

Nicodemus proposed BSSRDF (bidirectional surface scattering in 1977 Distribution function two-way dispersion surface reflectivity distribution function) theoretical, it can describe light on surface any two Transmittance process between point, Accurate Model Subsurface Scattering.

dL₀(x₀,ω₀)=f_bssrdf(x_i,ω_i；x₀,ω₀)dφ_i(x_i,ω_i)

Wherein, L₀(x₀,ω₀) it is x₀Point direction is ω₀Emergent radiation degree, φ_i(x_i,ω_i) it is from ω_iDirection is incided x_iLuminous flux.

Applying at present more in Rendering algorithms is the simple version BRDF (Bidirectional of BSSRDF Reflectance Distribution Function bidirectional reflectance distribution function), it is assumed that light is incident and outgoing is same Point, i.e. x₀=x_i, but this version can not render the visual effect of translucent material and quality is stiff does not have vivid effect.

In recent years, experts and scholars have obtained certain achievement in research in terms of subsurface Rendering algorithms:

Can accurately but time-consumingly simulate Subsurface Scattering by solving total radiation equation of transfer, but only have minority scholar at present Use the method, the photon mapping method such as Dorsey, simulate full Subsurface Scattering to render colorful pebble.Pharr and Hanrahan utilizes scattering function to simulate Subsurface Scattering.Although these methods solving total radiation equation of transfer can simulate institute There is Subsurface Scattering visual effect, but computation complexity is abnormal high.

Calendar year 2001, Jensen et al., on the basis of supposing that Subsurface Scattering surface is infinitely great half-plane, simplifies unrestrained Scattering model, it is proposed that illumination model in BSSRDF is decomposed into single Subsurface Scattering and repeatedly by the method for dipole light source approximation Subsurface Scattering, and row approximation when employing the method for dipole light source in repeatedly Subsurface Scattering, compare Monte Carlo can Obtain rendering speed faster.Simon in 2004 et al. observes laser and propagates in the liquid of full scattering and inspired, and carries Go out point spread function (point spread function) and combined optical attenuation volume pyramid, having further increased and render Speed, but spent internal memory is more than nine times of non-pyramid Rendering algorithms and once light source attributes changes, and needs pretreatment big Amount data, operand substantially increases, and Caton phenomenon occurs.

In sum, for the phenomenon consuming internal memory time-consuming in Subsurface Scattering Rendering algorithms, the present invention provides a kind of not Energy real-time rendering when changing light source attributes, only than the wash with watercolours rendering nontransparent material method extra consumption triploid size of data internal memory Dyeing method.

Summary of the invention

For the deficiencies in the prior art, an object of the present invention is to solve in Subsurface Scattering rendering intent, time-consumingly The problems such as consumption internal memory, it is proposed that the body rendering intent of a kind of translucent material, it is based on nontransparent Rendering algorithms, is not changing In the case of light source attributes, the Subsurface Scattering side of rendering of energy real-time rendering and only extra consumption triploid size of data internal memory Method.

To achieve these goals, the present invention adopts the following technical scheme that:

A kind of body rendering intent of translucent material, it comprises the following steps:

Step 1, acquisition light source attributes data and the material parameters data of translucent module body, described light source attributes includes light Source position L_p, light color L_cr、L_cgAnd L_cb；Described material parameters includes absorptance σ_a, scattering coefficient σ_sAnd phase functionWherein,For luminous energy incident direction,For luminous energy exit direction；

Step 2, to each light color spectrum, in conjunction with the material parameters data of light source attributes data and translucent module body Generate body light quantity model LV respectively_r、LV_gAnd LV_b:

${\mathrm{LV}}_j(x,y,z)=\frac{({\mathrm{\σ}}_{sj}+{\mathrm{\σ}}_{aj}-{\mathrm{\σ}}_{sj}\mathrm{\ϵ})L_{cj}}{d}{e}^{-d\sqrt{{\mathrm{\σ}}_{aj}{\mathrm{\σ}}_{sj}}}---\left(1\right)$

Wherein, j is r, g, b thrin, and d is coordinate x, the accumulation voxel value of y, z to light source direction；ε is scattering angle Mean cosine and:

$\mathrm{\ϵ}={\∫}_{4\mathrm{\π}}(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})p(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})d{\stackrel{\‾}{\mathrm{\ω}}}^{\′}---\left(2\right)$

Step 3, tradition light projecting algorithm on the basis of, add the impact on sampled point s of the sampled point surrounding voxels, The formula that renders to the translucent module body of monochromatic light:

$I\left(s_i\right)=I\left(s_{i-1}\right)T(s_{i-1},s_i)+{\∫}_{s_{i-1}}^{s_i}{\∫}_{2\mathrm{\π}}G(X,s)q\left(X\right)T(s,s_i)dXds---\left(3\right)$

Wherein, I (s_i) represent at sampled point s_iThe brightness at place, sampled point s_iRepresent that light source is by translucent module body Ith sample point；X represent be perpendicular to sight line and with the surrounding voxels on sampled point s intersecting plane, G (X, s) represent with sampling With the dimensional Gaussian distribution that physical depth D (d) is relevant centered by some s, D (d)=γ d, γ are the constant relevant to material；q(X) For light value at voxel X around in body light quantity model；T(s_i-1,s_i) represent sampled point s_i-1To sampled point s_iAccumulation transparent Degree and:

$T(s_{i-1},s_i)=e^{-\mathrm{\τ}(s_{i-1},s_i)}=e^{-{\∫}_{s_{i-1}}^{s_i}{\mathrm{\σ}}_a\left(s\right)ds}---\left(4\right)$

Wherein: τ (s_i-1,s_i) it is sampled point s_i-1To sampled point s_iBetween optical thickness, σ_aS () is that monochromatic light is at s point Absorptance value.

Described dimensional Gaussian distribution G (X, computational methods s) are:

G (X, s)=Gu_(σ,r)(|X-s|) (5)

Wherein: | X-s | is the surrounding voxels X vectorial coordinate from sampled point s；Gu_(σ,r)For the value of two-dimensional Gaussian kernel, the σ side of being Difference, r is radius, and:

σ=D (d) γ₁ (6)

R=D (d) γ₂ (7)

Wherein, γ₁For variance with change in depth multiple, γ₂For radius with change in depth multiple, γ₁And γ₂It is more than 1 Constant.

To achieve these goals, the present invention adopts the following technical scheme that:

A kind of body rendering system of translucent material, comprising:

Data collection module, for obtaining the material parameters data of light source attributes data and translucent module body, described light Source attribute includes light source position L_p, light color L_cr、L_cgAnd L_cb；Described material parameters includes absorptance σ_a, scattering coefficient σ_sWith Phase functionWherein,For luminous energy incident direction,For luminous energy exit direction；

Body light quantity model generation unit, for each light color spectrum, in conjunction with light source attributes data and translucent mould The material parameters data of block generate body light quantity model LV respectively_r、LV_gAnd LV_b:

${\mathrm{LV}}_j(x,y,z)=\frac{({\mathrm{\σ}}_{sj}+{\mathrm{\σ}}_{aj}-{\mathrm{\σ}}_{sj}\mathrm{\ϵ})L_{cj}}{d}{e}^{-d\sqrt{{\mathrm{\σ}}_{aj}{\mathrm{\σ}}_{sj}}}---\left(8\right)$

Wherein, j is r, g, b thrin, and d is coordinate x, the accumulation voxel value of y, z to light source direction；ε is scattering angle Mean cosine and:

$\mathrm{\ϵ}={\∫}_{4\mathrm{\π}}(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})p(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})d{\stackrel{\‾}{\mathrm{\ω}}}^{\′}---\left(9\right)$

Rendering unit, on the basis of tradition light projecting algorithm, adds sampled point surrounding voxels X to sampled point s Impact, obtain the formula that renders of the translucent module body of monochromatic light:

$I\left(s_i\right)=I\left(s_{i-1}\right)T(s_{i-1},s_i)+{\∫}_{s_{i-1}}^{s_i}{\∫}_{2\mathrm{\π}}G(X,s)q\left(X\right)T(s,s_i)dXds---\left(10\right)$

Wherein, I (s_i) represent at sampled point s_iThe brightness at place, sampled point s_iRepresent that light source is by translucent module body Ith sample point；X represent be perpendicular to sight line and with the surrounding voxels on sampled point s intersecting plane, G (X, s) represent with sampling With the dimensional Gaussian distribution that physical depth D (d) is relevant centered by some s, D (d)=γ d, γ are the constant relevant to material；q(X) For light value at voxel X around in body light quantity model；T(s_i-1,s_i) represent sampled point s_i-1To sampled point s_iAccumulation transparent Degree and:

$T(s_{i-1},s_i)=e^{-\mathrm{\τ}(s_{i-1},s_i)}=e^{-{\∫}_{s_{i-1}}^{s_i}{\mathrm{\σ}}_a\left(s\right)ds}---\left(11\right)$

Wherein: τ (s_i-1,s_i) it is sampled point s_i-1To sampled point s_iBetween optical thickness, σ_aS () is that monochromatic light is at s point Absorptance value.

Described dimensional Gaussian distribution G (X, computational methods s) are:

G (X, s)=Gu_(σ,r)(|X-s|) (12)

Wherein: | X-s | is the surrounding voxels X vectorial coordinate from sampled point s；Gu_(σ,r)For the value of two-dimensional Gaussian kernel, the σ side of being Difference, r is radius, and:

σ=D (d) γ₁ (13)

R=D (d) γ₂ (14)

Wherein, γ₁For variance with change in depth multiple, γ₂For radius with change in depth multiple, γ₁And γ₂It is more than 1 Constant.

GPGPU is write at GPU (Graphics Processing Unit graphic process unit) end according to formula (10) (General Purpose GPU general-purpose computations graphic process unit) programs render final image, finally reaches from GPU video memory CPU host memory refreshes for interface, and display renders image.

Preferably, described dimensional Gaussian distribution G (X, computational methods s) are:

G (X, s)=Gu_(σ,_r)(|X-s|) (12)

Wherein: | X-s | is the surrounding voxels X vectorial coordinate from sampled point s；Gu_(σ,r)For the value of two-dimensional Gaussian kernel, the σ side of being Difference, r is radius, and:

σ=D (d) γ₁ (13)

R=D (d) γ₂ (14)

Wherein, γ₁For variance with change in depth multiple, γ₂For radius with change in depth multiple, γ₁And γ₂It is more than 1 Constant.

The body rendering intent of the translucent material that the present invention illustrates and system, it has the beneficial effects that: throw at tradition light Expend a small amount of extra memory on the basis of penetrating three-dimensional rendering algorithm and realize real-time Subsurface Scattering simulation, solve existing mould very well Intend the time-consuming problem taking internal memory of Subsurface Scattering.

Accompanying drawing explanation

Fig. 1 is the flow chart of the body rendering intent of translucent material of the present invention；

Fig. 2 is tradition light projection three-dimensional rendering algorithm schematic diagram；

Fig. 3 is that light projecting algorithm color adds up schematic diagram；

Fig. 4 is the impact on sampled point of the sampled point surrounding voxels；

Fig. 5 is the change schematic diagram increasing radius along with the degree of depth；

Fig. 6 is the structural representation of integration plane；

Fig. 7 is gaussian kernel visualized graphs；

Fig. 8 is brain fMRI body rendering effect figure；

Fig. 9 is the block diagram of the body rendering system of translucent material of the present invention.

Detailed description of the invention

It is next with specific embodiment below in conjunction with the accompanying drawings that the invention will be further described.

Embodiment one

The body rendering intent of a kind of translucent material, is used for solving to simulate Subsurface Scattering time-consuming expense memory problem at present, Refer to shown in Fig. 1, it comprises the following steps:

S10, acquisition light source attributes and module body material parameters, light source attributes includes: light source position L_p, light color L_cr、L_cg And L_cb(being followed successively by HONGGUANG color, green glow color and blue light color)；Material parameters includes: absorptance σ_a, scattering coefficient σ_sWith Phase functionWherein absorptance σ_a=σ_a(r, g, b), it includes HONGGUANG color absorption factor sigma_ar, green glow color absorption Factor sigma_agAnd blue light color absorptance σ_ab, similarly scattering coefficient σ_s=σ_s(r, g, b), it includes that the scattering of HONGGUANG color is Number σ_sr, green glow color scattering coefficient σ_sgAnd blue light color scattering coefficient σ_sb。

Mean cosine ε of definition scattering angle is

$\mathrm{\ϵ}={\∫}_{4\mathrm{\π}}(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})p(\stackrel{\‾}{\mathrm{\ω}}-{\stackrel{\‾}{\mathrm{\ω}}}^{\′})d{\stackrel{\‾}{\mathrm{\ω}}}^{\′}---(1-1)$

S20, to each color spectrum, light source position and material parameters that integrating step S10 obtains generate body light quantity model LV_r、LV_gAnd LV_b,

${\mathrm{LV}}_j(x,y,z)=\frac{({\mathrm{\σ}}_{sj}+{\mathrm{\σ}}_{aj}-{\mathrm{\σ}}_{sj}\mathrm{\ϵ})L_{cj}}{d}{e}^{-d\sqrt{{\mathrm{\σ}}_{aj}{\mathrm{\σ}}_{sj}}}---(2-1)$

Wherein, j is j, g, b thrin；D is coordinate x, the accumulation voxel value of y, z to light source direction, such as, when j is r, (2-1) formula then makes into:It is the body of HONGGUANG color spectrum Light quantity model, in like manner, the body light quantity model LV of the most corresponding green glow color spectrum when j is g or b_gBody with blue light color spectrum Light quantity model LV_b。

S30, rendering, traditional light projecting algorithm (i.e. light projection three-dimensional rendering algorithm) is as shown in Figure 2.

Light projection method is direct volume drawing algorithm based on image sequence.From each pixel of image, along fixing A light is launched in direction (typically direction of visual lines), the whole image sequence of light traverses, and in this process, to image sequence Row carry out sampling and obtain colouring information, color value are added up, until light traverses is whole according to light absorption model simultaneously Image sequence, the color value obtained is exactly the color rendering image, and all pixels on last traverses screen are just specified Object two dimensional image on screen.

As it is shown on figure 3, existing color (monochromatic light) totalization formula is:

$I\left(s_i\right)=I\left(s_{i-1}\right)T(s_{i-1},s_i)+{\∫}_{s_{i-1}}^{s_i}q\left(s\right)T(s,s_i)ds---(3-1)$

Wherein,Q (s) is voxel radiation value, σ_aS () is monochromatic light (three bases One in color) or other a certain light in the absorptance value of s point, three primary colours can independently with the absorptance of oneself, Scattering coefficient using formula (3-1) calculates I_r(s_i)、I_g(s_i) and I_b(s_i), finally combine formation colored rendering figure together.

Because tradition Projection algorithm does not considers, along sight line sampled point surrounding voxels, the contribution of last image does not the most consider time table Area scattering, therefore generates image stiff the most true to nature.

The present invention, on the basis of tradition Projection algorithm, adds the impact on sampled point of the sampled point surrounding voxels, such as Fig. 4 institute Show, after i.e. obtaining amendment, can make the formula that renders of the more life-like translucent module body of monochromatic light of generation image:

$z\left(s_i\right)=z\left(s_{i-1}\right)T(s_{i-1},s_i)+{\∫}_{s_{i-1}}^{s_i}{\∫}_{2\mathrm{\π}}G(X,s)q\left(X\right)T(S,S_i)dXds---(3-2)$

Wherein, X (being defined as surrounding voxels) for be perpendicular to sight line and with the point on sampled point intersecting plane, it is known that normal and On normal a bit, geometry knowledge is used can conveniently to try to achieve intersecting plane and at this planar integral.R and θ is respectively the pole of X point Radius and polar angle, dX=drd θ, 0≤θ≤2 π.(X is s) with the dimensional Gaussian that degree of depth D (d) is relevant divides centered by sampled point to G Cloth, wherein D is physical depth, and for uniform material, it is the fixing multiple of accumulation voxel value d, i.e. D=γ d, wherein γ and material The constant of qualitative correlation.The general the deepest variance of the degree of depth is the biggest, and q (X) is the light quantity in the body light quantity model that step S20 generates at X point Value.

It is demonstrated experimentally that in calculating the surrounding voxels light quantity contribution to sampled point S, the voxel light quantity that distance sample is the most remote Contributing the least, therefore the present invention is distributed in order to the dimensional Gaussian centered by sampled point S and simulates the contribution of nearly voxel greatly, remote voxel tribute Offer little situation, if Fig. 5 radial direction stepping is R_step；Experiment demonstrates again that, along with the degree of depth increase dispersion effect also with Enhancing, voxel contribution margin also reaches unanimity, therefore the present invention with the radius r (i.e. Gauss window size) increased with the degree of depth and with The Gauss variances sigma that the degree of depth increases simulates this effect, if Fig. 5 direction of visual lines stepping is S_step.

Variance and radius determine value Gu of two-dimensional Gaussian kernel_(σ,r), the formula that variances sigma and radius r change with degree of depth D is:

σ=D γ₁ (3-3)

R=D γ₂ (3-4)

Wherein, γ₁And γ₂Identical material is greater than two constants of 1, bigger for relatively transparent material value.As Fig. 6 is high This core is about centrosymmetry, and therefore can set from centre coordinate | X-s | is independent variable, then (X s) counts the G in formula (3-2) Calculation formula is

G (X, s)=Gu_(σ,r)(|X-s|) (3-⁵)

Wherein, | X-s | is the X vectorial coordinate from sampled point s, and along with X is away from sampled point s, (X s) reduces, s weight G_xWith s_yIt is abscissa and the vertical coordinate of sampled point s respectively, because the definition polar form (r, θ) of X point, so needing to be converted Become plane coordinates, the plane coordinates parameter of | the X-s | that then reentries, it may be assumed that

| X-s |=(r cos θ-s_x,r·sinθ-s_y) (3-6)

Below as a example by brain fMRI body renders, said process is explained in detail and illustrates.

First, obtaining light source and material parameters, arranging light source position is and x, y and z axes angle 45 degree, a length of 500 Vector on；The material parameters arranging human body skin is: σ_s(0.74,0.88,1.01), σ_a(0.032,0.17,0.48).Wherein Each component is scattering or the absorption parameter of each color, and i.e. 0.74 is the scattering coefficient to r (red), can be expressed as σ_sr, in like manner, 0.88 is the scattering coefficient σ to g (green)_sg, 1.01 is the scattering coefficient σ to b (blue)_sb。

2. can generate the light quantity body Model of redgreenblue after getting parms, so-called light quantity body Model is exactly that light enters material Light radiation in plastid, generates three individual data items according to equation below.

${\mathrm{LV}}_j(x,y,z)=\frac{({\mathrm{\σ}}_{sj}+{\mathrm{\σ}}_{aj}-{\mathrm{\σ}}_{sj}\mathrm{\ϵ})L_{cj}}{d}{e}^{-d\sqrt{{\mathrm{\σ}}_{aj}{\mathrm{\σ}}_{sj}}}$

Wherein as described in technical scheme, j is r, g, b thrin；D is coordinate x, the accumulation voxel of y, z to light source direction Value, light color L_cr、L_cg、L_cb, material absorptance σ_a, scattering coefficient σ_sMean cosine ε with scattering angle.Do not changing light source When position or other attribute, this light body-measure data only needs to calculate once.

3. in the case of light source position and material parameters are the most immovable, three light quantity body Models can keep constant and Directly carry out rendering step.Render stepping and be typically chosen 0.5 in the case of not producing moire effect, every stepping is once sampled Point, obtains light quantity body Model value about, and the stepping selecting radius is identical with sampled point number of steps, i.e. R_step in Fig. 5 with S_step is identical, and maximum radius Max_d is the function of depth capacity Max_d:

$Max\_r=\frac{255}{{\mathrm{\σ}}_s}Max\_d$

Wherein depth capacity is material properties, represents the depth capacity that light can arrive, can arrange in implementing Along the maximum iteration time of light sampling, it is typically chosen 500 steps.

4. render the equation of improvement according to conventional bulk:

$I\left(s_i\right)=I\left(s_{i-1}\right)T(s_{i-1},s_i)+{\∫}_{s_{i-1}}^{s_i}{\∫}_{2\mathrm{\π}}G(X,s)q\left(X\right)T(s,s_i)dXds$

Write GPGPU programs render final image at GPU end, from GPU video memory, finally reach CPU host memory for interface Refreshing, display renders image.Wherein, X be perpendicular to sight line and with the point on sampled point intersecting plane, (X is s) with sampled point to G Centered by with degree of depth D (d) relevant dimensional Gaussian distribution, the biggest variance of the general degree of depth is the biggest, and q (X) is at X in body light quantity model The light value of point.Such as, radius is 3, and variance is that gaussian kernel when 1.5 is

Its integration plane graph and visualized graphs are the most as shown in Figure 6 and Figure 7.

5. different material properties values is set

In being embodied as, providing the interface arranging material properties value on interface, user can arrange accumulation voxel value d's Fixing multiple γ, variance are with change in depth multiple γ₁, radius is with change in depth multiple γ₂And depth capacity, maximum iteration time Observe the Same Way rendering effect to unlike material.Generally the parameter of this method is γ=1, i.e. accumulation voxel Value d is physical depth D；γ₁And γ₂It is respectively 2.3 and 1.8, thus obtains brain fMRI body rendering effect figure, such as Fig. 8 institute Show.

Embodiment two

Embodiment two provides the body rendering system of a kind of translucent material, refer to shown in Fig. 9, and it includes data collection list Unit 10, body light quantity model generation unit 20 and rendering unit 30, they realize step S10, S20 and S30 in embodiment one respectively Flow process, repeat no more here.

The above, be only present pre-ferred embodiments, not impose any restrictions the technical scope of the present invention, therefore Every any trickle amendment, equivalent variations and modification made above example according to the technical spirit of the present invention, the most still belongs to In the range of technical solution of the present invention.

Claims (4)

1. the body rendering intent of a translucent material, it is characterised in that it comprises the following steps:

Step 1, acquisition light source attributes data and the material parameters data of translucent module body, described light source attributes includes light source position Put L_p, light color L_cr、L_cgAnd L_cb；Described material parameters includes absorptance σ_a, scattering coefficient σ_sAnd phase functionIts In,For luminous energy incident direction,For luminous energy exit direction；

Step 2, to each light color spectrum, in conjunction with light source attributes data and translucent module body material parameters data respectively Generate body light quantity model LV_r、LV_gAnd LV_b:

${\mathrm{LV}}_j(x,y,z)=\frac{({\mathrm{\σ}}_{sj}+{\mathrm{\σ}}_{aj}-{\mathrm{\σ}}_{sj}\mathrm{\ϵ})L_{cj}}{d}{e}^{-d\sqrt{{\mathrm{\σ}}_{aj}{\mathrm{\σ}}_{sj}}}---\left(1\right)$

Wherein, j is r, g, b thrin, and d is coordinate x, the accumulation voxel value of y, z to light source direction；ε is the flat of scattering angle All cosine and:

$\mathrm{\ϵ}={\∫}_{4\mathrm{\π}}(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})p(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})d{\stackrel{\‾}{\mathrm{\ω}}}^{\′}---\left(2\right)$

Step 3, tradition light projecting algorithm on the basis of, add the impact on sampled point s of the sampled point surrounding voxels, obtain list The translucent module body of coloured light render formula:

$I\left(s_i\right)=I\left(s_{i-1}\right)T(s_{i-1},s_i)+{\∫}_{s_{i-1}}^{s_i}{\∫}_{2\mathrm{\π}}G(X,s)q\left(X\right)T(s,s_i)dXds---\left(3\right)$

Wherein, I (s_i) represent at sampled point s_iThe brightness at place, sampled point s_iRepresent that light source is by i-th on translucent module body Individual sampled point；X represent be perpendicular to sight line and with the surrounding voxels on sampled point s intersecting plane, G (X, s) represent with sampled point s be Center is distributed with the dimensional Gaussian that physical depth D (d) is relevant, and D (d)=γ d, γ are the constant relevant to material；Q (X) is body Light value at voxel X around in light quantity model；T(s_i-1,s_i) represent sampled point s_i-1To sampled point s_iAccumulation transparency And:

$T(s_{i-1},s_i)=e^{-\mathrm{\τ}(s_{i-1},s_i)}=e^{-{\∫}_{s_{i-1}}^{s_i}{\mathrm{\σ}}_a\left(s\right)ds}---\left(4\right)$

Wherein: τ (s_i-1,s_i) it is sampled point s_i-1To sampled point s_iBetween optical thickness, σ_aS () is the monochromatic light absorption at s point Coefficient value.

The body rendering intent of translucent material the most according to claim 1, it is characterised in that described dimensional Gaussian distribution G (X, computational methods s) are:

G (X, s)=Gu_(σ,r)(|X-s|) (5)

Wherein: | X-s | is the surrounding voxels X vectorial coordinate from sampled point s；Gu_(σ,r)For the value of two-dimensional Gaussian kernel, σ is variance, r For radius, and:

σ=D (d) γ₁ (6)

R=D (d) γ₂ (7)

Wherein, γ₁For variance with change in depth multiple, γ₂For radius with change in depth multiple, γ₁And γ₂Be more than 1 is normal Number.

3. the body rendering system of a translucent material, it is characterised in that comprising:

Data collection module, for obtaining the material parameters data of light source attributes data and translucent module body, described light source belongs to Property includes light source position L_p, light color L_cr、L_cgAnd L_cb；Described material parameters includes absorptance σ_a, scattering coefficient σ_sWith phase letter NumberWherein,For luminous energy incident direction,For luminous energy exit direction；

Body light quantity model generation unit, for each light color spectrum, in conjunction with light source attributes data and translucent module body Material parameters data generate body light quantity model LV respectively_r、LV_gAnd LV_b:

${\mathrm{LV}}_j(x,y,z)=\frac{({\mathrm{\σ}}_{sj}+{\mathrm{\σ}}_{aj}-{\mathrm{\σ}}_{sj}\mathrm{\ϵ})L_{cj}}{d}{e}^{-d\sqrt{{\mathrm{\σ}}_{aj}{\mathrm{\σ}}_{sj}}}---\left(8\right)$

Wherein, j is r, g, b thrin, and d is coordinate x, the accumulation voxel value of y, z to light source direction；ε is the flat of scattering angle All cosine and:

$\mathrm{\ϵ}={\∫}_{4\mathrm{\π}}(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})=(\stackrel{\‾}{\mathrm{\ω}},{\stackrel{\‾}{\mathrm{\ω}}}^{\′})d{\stackrel{\‾}{\mathrm{\ω}}}^{\′}---\left(9\right)$

Rendering unit, for, on the basis of tradition light projecting algorithm, adding the sampled point surrounding voxels X shadow to sampled point s Ring, obtain the formula that renders of the translucent module body of monochromatic light:

$I\left(s_i\right)=I\left(s_{i-1}\right)T(s_{i-1},s_i)+{\∫}_{s_{i-1}}^{s_i}{\∫}_{2\mathrm{\π}}G(X,s)q\left(X\right)T(s,s_i)dXds---\left(10\right)$

Wherein, I (s_i) represent at sampled point s_iThe brightness at place, sampled point s_iRepresent that light source is by i-th on translucent module body Individual sampled point；X represent be perpendicular to sight line and with the surrounding voxels on sampled point s intersecting plane, G (X, s) represent with sampled point s be Center is distributed with the dimensional Gaussian that physical depth D (d) is relevant, and D (d)=γ d, γ are the constant relevant to material；Q (X) is body Light value at voxel X around in light quantity model；T(s_i-1,s_i) represent sampled point s_i-1To sampled point s_iAccumulation transparency And:

$T(s_{i-1},s_i)=e^{-\mathrm{\τ}(s_{i-1},s_i)}=e^{-{\∫}_{s_{i-1}}^{s_i}{\mathrm{\σ}}_a\left(s\right)ds}---\left(11\right)$

Wherein: τ (s_i-1,s_i) it is sampled point s_i-1To sampled point s_iBetween optical thickness, σ_aS () is the monochromatic light absorption at s point Coefficient value.

The body rendering system of translucent material the most according to claim 3, it is characterised in that described dimensional Gaussian distribution G (X, computational methods s) are:

G (X, s)=Gu_(σ,r)(|X-s|) (12)

Wherein: | X-s | is the surrounding voxels X vectorial coordinate from sampled point s；Gu_(σ,r)For the value of two-dimensional Gaussian kernel, σ is variance, r For radius, and:

σ=D (d) γ₁ (13)

R=D (d) γ₂ (14)

Wherein, γ₁For variance with change in depth multiple, γ₂For radius with change in depth multiple, γ₁And γ₂Be more than 1 is normal Number.