Ambient occlusion and edge cueing to enhance real time molecular visualization

IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS,VOL.12,NO.5,SEPTEMBER/OCTOBER2006

Ambient Occlusion and Edge Cueing

to Enhance Real Time Molecular Visualization

Marco Tarini,Paolo Cignoni,and Claudio Montani

Abstract—The paper presents a set of combined techniques to enhance the real-time visualization of simple or complex molecules (up to order of106atoms)space?ll mode.The proposed approach includes an innovative technique for ef?cient computation and storage of ambient occlusion terms,a small set of GPU accelerated procedural impostors for space-?ll and ball-and-stick rendering, and novel edge-cueing techniques.As a result,the user’s understanding of the three-dimensional structure under inspection is strongly increased(even for still images),while the rendering still occurs in real time.

!

1I NTRODUCTION

Interactive protein visualization is an important application with harder requirements every year:public databases like RCBS Protein Data Bank[1]are storing a large number of molecular structures of ever-increasing complexity.It is essential to be able to visualize the shape of these proteins in an interactive,meaningful and insightful way so that users can correctly understand the three-dimensional struc-ture of these shapes.

Many software systems,such as for example RasMol[28],CN3D [6]or Chimera[5],are available to help scientists in this task.These systems offer various visualization modalities that map the inner struc-ture of the molecules into3D shapes according to an almost standard set of paradigms as Balls-and-Sticks,Space-Fill,Licorice,Ribbons and various kinds of accessibility surfaces.These approaches are able to describe local geometrical and chemical properties of the inspected structure and to provide insights on chemical traits of the molecule. An emerging problem is that,due to the growing size and complexity of the analyzed proteins,all of these visualization modes map proteins into three dimensional structures that,when rendered with standard local shading techniques,fail to produce a high-level comprehensible picture to the user.In other words,for complex molecules it is very dif?cult to perceive the overall3D structure of the protein from a sin-gle image.This problem is only diminished,but not solved,when other common visualization techniques,as ribbons,provide a higher level description of the structure.

From a rendering oriented point of view,the problem is closely re-lated to the use of local lighting models,since global illumination ef-fects are usually not available in a handy interactive way.Sometimes more sophisticated approaches are used by means of off-line tools like raytracers for generating high quality images for important presenta-tion cases(like journal or web covers);but these tools never aid the research during the interactive visualization process.

The most common(non-local)effects that are used for enhancing the rendering are cast shadows and depth cueing.Both of them are available in most of the well-known molecular visualization systems, but these approaches often fail to create an easily comprehensible im-age.For example,consider the molecule shown in Fig.7lower left: while it is clearly better than the plain rendering(top left),the shape of some portions of the protein is still very ambiguous.Other common ?Marco Tarini is with Universit`a dell’Insubria,Varese,Italy, E-mail:m.tarini@https://www.wendangku.net/doc/ec7244984.html,r.it.

?Paolo Cignoni is with I.S.T.I.-C.N.R,Pisa,Italy,

E-mail:p.cignoni@https://www.wendangku.net/doc/ec7244984.html,r.it

?Claudio Montani is with I.S.T.I.-C.N.R,Pisa,Italy,

E-mail:c.montani@https://www.wendangku.net/doc/ec7244984.html,r.it

Manuscript received31March2006;accepted1August2006;posted online6 November2006.

For information on obtaining reprints of this article,please send e-mail to: tvcg@https://www.wendangku.net/doc/ec7244984.html,.techniques adopted to overcome this problem are interactivity and the use of stereoscopic displays.

In recent years much of the research efforts on large protein vi-sualization has targeted ef?cient rendering of these large3D struc-tures,to achieve real interactivity.Various systems were presented with this purpose,both relying on existing3D API[4]or trying to exploit features of recent graphics hardware or multiresolution tech-niques[11,10].Importantly,while the focus of this paper is to raise the quality of the perception of the shapes through sophisticated shad-ing and edge cueing effects(summarized in Fig.1),the interactivity of the whole system is a fundamental feature of our approach:all pre-sented techniques work in realtime for very large proteins like the 1AON depicted in Fig.7(around60K atoms)and do not need any preprocessing phase of signi?cant length.

The main contributions of this paper can be summarized as follows:?the use of advanced shading techniques for enhancing the per-ception of the3D shape of large proteins;

?a novel technique for parameterizing the surface of molecules represented as SpaceFill or Ball and Sticks models(sec.3.3);

?a novel technique for ef?ciently computing ambient occlusion information for molecules(sec.4);

?an ef?cient approach for the rendering of molecular models using GPU based procedural textured impostors.(sec.3.1);

?two interactive techniques for enhancing edges in molecular ren-derings(sec.5).

2P REVIOUS WORK.

We aim to enhance the shape perception process during molecular vi-sualization by means of stylistic rendering techniques combined with a more sophisticated shading model.In other words,we are de?n-ing an illustrative visualization approach[31,8]for enhancing the information-effectiveness of the rendered images of molecules.

We adopt more a sophisticated shading model that approximates the light coming from an uniformly diffusing lighting environment. This was shown to be useful in[17],where the authors report the re-sults of perceptual experiments showing that depth discrimination un-der diffuse lighting is superior to that predicted by a classical sunny day/direct lighting model,and by a model in which perceived lumi-nance varies with depth.The inadequacy of local lighting models was already noted in[30],where the use of a vicinity shading,a variant of the obscurance term proposed in[32],was proposed to enhance the visualization of volumetric datasets.Similarly in[21]the accessibil-ity shading approach was introduced,where the geometric local(and global)accessibility of a point is used to modify the Lambertian shad-ing of the surface in order to darken deep,dif?cultly accessible areas. Enhancing edges and silhouette As was shown in[26,9,20, 14]?nding and displaying silhouette edges is an important task that is used by illustrative rendering approaches to improve the readability of a3D scene.Many different techniques have been proposed and a good survey on this matter can be found in[13].Some of the proposed

IEEE TRANSACTIONS ON VISUALIZA TION AND COMPUTER GRAPHICS,VOL.12,NO.5,SEPTEMBER/OCTOBER

2006 Fig.1.Good molecular rendering can be more informative,clearer,and more capable of communicating a shape when it is the combined result of several effects and techniques.Top row:base atom colors,direct Lambertian and Phong illumination,self-shadowing,depth cueing.Bottom row (the topics of this paper):ambient occlusion,depth-aware halos,depth-revealing contour lines,and intersection-revealing contour lines.

approaches,like[26],work directly in image space,by processing the rendered depth buffer of the scene;others works on the mesh repre-sentation?nding edges of the mesh that are part of the silhouette for the current viewpoint.Other authors exploit graphics hardware ren-dering features to perform this task,such as the z-offsetting technique presented in[25]that produces borders by rendering front and back faces of a scene with a different z-offset.Recent approaches,like ours, exploit the programmability of the graphics hardware to implement directly on the GPU most of these approaches.A more detailed com-parison between our approaches for enhancing the edges of a molecule and the existing approaches is reported in section5.

Enhancing the ambient term In local shading models the effect light which does not come directly from the primary light source has to be approximated.Otherwise,the portion of the scene which is not directly lit will come out entirely dark.Even without resorting to more correct(and complex)global illumination solutions,shortcuts are pos-sible.The commonest and cheapest solution[23]is to use a simple per-scene constant term,but this approach leads to a notable?atness in the portions of the scene which are not directly lit.

The approach has been improved by explicitly computing for each point of the surface its accessibility value,which is the percentage of the hemisphere above each surface point not occluded by geometry [16].This useful technique is commonly known as ambient occlusion and it is used in many production environments to add an approxima-tion of the shadowing of diffuse objects lit with environment lighting. For example,ambient occlusion is precomputed in the interactive vi-sualization system described in[2].

Variants of the ambient occlusion term called obscurance have been introduced in[32,12],where the authors propose to exponentially weight the occlusion factor according to the distance of the occlud-ers in order to enhance the shadowing effects of near occluding sur-faces.In all of the above proposals,the computation of this extended ambient term is performed using a traditional ray-traced approach.In [27]graphics hardware is exploited to ef?ciently compute a per ver-tex ambient occlusion term.They render all geometry as seen from a light source direction into the depth buffer.Then,all vertices are rendered again as a point set.For each vertex an individual hardware based occlusion query is used to?nd vertices that passed the depth test and which are therefore visible for the considered light source;?nally a visibility matrix M is stored per vertex.A similar approach,based on the use of the GPU for the computation of the ambient occlusion term,has been proposed in[22],and extended to the computation of a ?rst bounce of the diffuse interre?ection of light in[3].Our approach for computing the ambient occlusion term is somewhat similar to the above techniques,but it exploits a different parametrization and access strategy for storing the computed results relying on the procedural na-ture of the molecular datasets;moreover,we add two signi?cant opti-mizations:one exploiting the structure of molecular shapes,the other extendible to the general case.

As a?nal note,’accessibility’itself is quite an important concept in chemistry:introduced by Lee and Richards[18]it is used to determine which parts of a given molecule are accessible to(the spherical atoms of)another molecule.Note that this paper does not discuss the direct visualization of this chemical property but focuses on the improve-ments on3D shape perception triggered by(approximate)non-local shading models and other illustrative rendering techniques.

2.1Ambient Occlusion De?nitions

Let us consider a point p on the surface with surface normal n p.Ac-cording to[15]we can de?ne the irradiance,E,arriving at p as:

E(p)=

?

n p·ωL(ω)dω(1) where L(ω)is a scalar with magnitude equal to the radiance arriving from directionω,and?is the set of directions above the surface,i.e. the direction for which n p·ω>0.This can be approximately eval-uated by discretizing the domain?into k sectorsωi with a possibly uniform solid angle measure|ωi|,and,for each sector,evaluating the radiance L only for a sample directionωi:

E(p)=

k

∑

i=1

n p·ωi L(ωi)|ωi|(2)

The above equation becomes simpler if we consider a uniform light-ing environment(where light comes uniformly from every direction, as under a cloudy sky).In this case,if we discard diffuse interre?ec-tion effects and therefore we take into account only direct lighting, L(ω)can be substituted by a simple binary function O(ω)valued0 if the ray shoot from p alongωintersects our surface(and therefore the light coming from the sky is obscured)and1otherwise.The re-sult can be considered a simple?rst order approximation of the whole rendering equation.

With the assumption of a uniform sampling of theωdirections,

E(p)=

1

4π

k

∑

i=1

n p·ωi O(ωi)(3)

3E FFICIENT R ENDERING OF B ALL AND STICK AND S PACE

F ILL

In our scenario the scene is composed by two simple types of prim-itives:cylinders(side area only)and spheres.We use2D impostors

TARINI,ET AL.:AMBIENT OCCLUSION AND EDGE CUEING FOR MOLECULAR VISUALIZA TION

for both primitives.One impostor is dedicated to each occurrence of cylinder or sphere in the scene,and it is rendered as a2D rectangu-lar quad in the viewing plane which encapsulates the projection of the primitive it stands for(see Fig.5).

The rationale is that impostors can be made much more rendering-ef?cient than a triangular tessellations,both during pre-computation of global illumination and in the?nal rendering.This is critical because we need to target large(>50k atoms)organic molecules as well.This is also memory-ef?cient,as the surface of the molecule is succinctly de?ned in an implicit way.The result is visually better,as the triangle tessellations are just linear approximations bound to produce shading and intersection artifacts.

Another advantage of the impostor approach is that,working on im-age space,a number of rendering effects(in particular border-oriented ones)become straightforward and computation-friendly(see Sec.5).

However,this approach poses the problem of how to store a signal, de?ned over the implicit surface,to record the precomputed ambient occlusion terms.This problem is addressed in Sec.3.3.

3.1Procedural impostors

We need our impostors to be z-,normal-,α-and u,v-mapped.For a given view,each2D point q inside an impostor represents a3D point p on the primitive P(the point visible through it),and has the following attributes:

?the Boolean membership of q inside the projection of P—in order to discard the corresponding fragment otherwise;

?the normal p n of p—in order to apply direct illumination;

?the depth z of the fragment—in order to correctly compute the intersections between primitives,and also to compute shadow-maps;

?the texture position u,v for the corresponding visible3D point on the primitive—in order to access any attribute previously stored for p.

The last point is crucial,because we need to store per-position infor-mation for our primitives(in particular,a value for ambient occlusion, see Sec.4).In practice we are resorting to a global2D parameteri-zation of the entire surface of the molecule,intended as the set of the surfaces of all its primitives(see Sec.3.3).

A solution could be to store the listed attributes in a set of?xed textures accessed for each fragment.This could impact perfor-mance because of the additional texture bandwidth consumption,and would cause aliasing problems(especially around borders,as pres-ence of semitransparent interpolated texels makes the rendering sort-dependent).On the contrary,our impostors are procedural,meaning that all attributes are synthesized on the?y.This greatly helps aliasing problems,and also improves?exibility and adaptability.

3.2General schema

For each impostor,we send four vertices with appropriate values.

A vertex program is dedicated to expand the impostor around the processed primitive,aiming at producing the least number of frag-ments outside the impostor(but producing all the fragments relative to the front facing part of the primitive).The initial vertex position is projected and the impostor is then expanded in image space.This happens differently for the two primitives(see Sec.3.3).

Another objective of the vertex shader is to perform as many pre-computations as possible in order to minimize residual per-fragment workload(for this reason we prefer not to use the Point Sprites exten-sion).This general optimization technique is,in our case,particularly fruitful because our impostors represent large,high-level primitives, and the pixel-to-vertex ratio is accordingly larger than usual.

Constant per-primitive parameters(e.g.base atom color)are passed down unchanged by the vertex program to the fragment processor.

The fragment program computes and then processes the required ?elds,including membership,u?v texture position,depth,or light-ing(according to need of the current rendering mode and rendering pass).Depending on the rendering technique,the?nal values are writ-ten either in the current screen buffers or in intermediate textures for subsequent passes.3.3Parameterizing the surface of a molecule

Our visualization algorithm requires a data structure to store the com-puted ambient occlusion terms.This can be assumed to be a low fre-quency signal,i.e.to vary smoothly over the surface of the molecule. For the case of atoms,this signal is de?ned over spheres,so it would be possible to store it,in a conveniently compact way,as a small set of spherical harmonic coef?cients[29]per atom.However compact this representation would be,we choose a direct sampling represen-tation because it is more local in nature and therefore more ef?cient during rendering:only few samples need to be read and interpolated to reconstruct the signal in a speci?c location.

Now we need a way to store a sampling over spheres and cylin-ders.Since the surfaces are implicit,we resort to a specially formatted texture that is to be coherently accessed during the rendering of the impostors.

We assign to each instance of sphere or cylinder a unique rectangu-lar(non necessarily squared)patch of texture space.All patches have the same height so that they can be trivially packed in a single global texture.During any rendering that requires texture accesses,the2D offset of the patch,relative to the origin of the the global texture,is sent as an additional attribute.

Depending on the size of the molecule(number of atoms)the sizes of texture patches vary from as few as4to hundreds of texels per side; for example,for a molecule with around1K atoms32×32patches can be packed in a1024×1024texture,while the surface of up to64K atoms can be sampled into4×4large patches of a1024×1024texture (when many atoms are present,the radius of the molecule is probably very large as well,reducing atom average screen size and which means that fewer texels per atom will suf?ce).

For both kinds of primitive we de?ne a mapping M between every point on its area and a position inside the corresponding texture patch. The function M needs to be simple,as it will be computed both ways within the fragment shader:during scene rendering,given a point in-side the impostor q,we need to compute M(P(q))=(u,v)to?nd tex-ture coordinates for current fragment;during ambient occlusion com-putation(see later in sec4),we will need to compute M?1for each fragment.Naturally we seek functions M that exhibit low distortions, so that texture mapping artifacts are minimized.

We use one of two alternative parametrization schemas M s1and M s2 for spheres,and one schema M c for cylinders.

Spheres:parameterization The mapping M s1is a gnomonic projection over a cube,followed by packing of its6faces in a rectangu-lar patch with a2×3aspect ratio.Following[24],M s2is a gnomonic projection over an octahedron,which is unfolded into a square.The mapping M s2and its inverse are easier to compute,taking fewer op-erations in the fragment shader(in our implementation,9ARB low level fragment operation instead of16),whereas M s1presents a minor stretch energy([24]).Another advantage of M s2over M s1is that it requires less duplicated texels(see later in Sec3.5),so no choice fully dominates the other.This choice is orthogonal with the rest of any of the algorithm discussed here.

M s1and its inverse are well known.We report the exact version that we use for M s2,that goes from the surface the unit,origin centered sphere to the patch parameterized as the square[?1..+1]2(we will use a GPU friendly formulation,with the fewest possible cases):

M s2(x,y,z)=

(x

d

,y

d

)if z≤0,

(sign(x)(1?|y|

d

),sign(y)(1?|x|

d

))if z>0

(4) where d=|x|+|y|+|z|.The inverse,up to a normalization,is:

M?1

s2

(u,v)=

(u,v,h)if h≥0,

(sign(u)(1?|v|),sign(u)(1?|v|),h)if h<0

(5) where h=1?|u|?|v|

Note that the mapping M s1,being gnomonic,does not need its ar-gument to be normalized prior to use.

IEEE TRANSACTIONS ON VISUALIZA TION AND COMPUTER GRAPHICS,VOL.12,NO.5,SEPTEMBER/OCTOBER

2006 Fig.2.Applying a small2D texture map over a sphere impostor,using

a10×10octahedron mapping M s2.In order to show the behavior of

M s2the virtual sphere(impostor)is rotated from left to right by180o.For

illustration purposes bilinear interpolation is disabled and random values

are assigned to texels,so that they are clearly visible.Duplicated texels

are assigned to the same color(see later).Refer to?g.6to see the

used2D texture.

Spheres:impostors(see Fig.2).Four vertices are sent at the

position of the sphere center,distinguished by the value of a special

attribute(s,t)assigned respectively to(±1,±1)to designate each ver-

tex to one different corner of the impostor quad.

Each vertex is displaced in screen coordinates after projection to

produce a screen aligned quad.The displacement is given by(s·r·

S g,t·r·S g,),where r is the radius of the sphere(passed as an attribute)

and S g is the global scale factor,extracted once per frame from the

current view matrix(as the cubic root of its determinant)and stored in

a environment parameter.

Interpolated values for s and t are passed down to the rasterizer,

so each fragment inside the screen quad is produced with an as-

signed relative position(s,t)∈[?1..+1]2.We call this2D space

impostor space.Fragments with|(s,t)|>1are immediately dis-

carded.For every surviving fragment,the original z value is de-

creased by(1?|(s,t)|)?r and its normal in screen space is assigned to

n=(s,t,(1?|(s,t)|)).The normal is then transformed in object space

with n =A?1MV(n)(i.e.by a multiplication with the inverse—i.e.the

transpose—of the current model-view matrix),and the result is fed to

the chosen mapping(u,v)=M s(n ).Texture is fetched at the position

(u,v)plus the offset for the current patch.

Cylinders:parametrization Let us consider the normalized,z-

axis aligned cylinder centered in the origin de?ned as the set of points

{(x,y,z)|x2+y2=1,x∈[?1..+1]}.

The side area of cylinders is developable,so it would be natu-

rally parameterizedwith the zero-stretch parametrization M(x,y,z)=

(atan2(x,y)/π,z).

However computing it in the fragment shader(either direction)

would be very demanding,since it requires trigonometric functions

(direct and inverse)which are not supported in the typical GPU.These

functions could be either approximated(e.g.with Taylor formulas),

which is time consuming,or sampled from a1D texture,which re-

quires additional texture accesses.We prefer a cheaper alternative,

consisting in the adoption of a simpli?ed mapping M c,which is a pro-

jection over a square-based prism and is de?ned by:

M c(x,y,z)=

(y

2(|x|+|y|)

?0.5,z)if x≥0,

?(y

2(|x|+|y|)

+0.5,z)if x<0

(6)

Note:the prism sides are de?ned on the|x|±|y|=±1planes rather than the|x|=±1and|y|=±1to simplify the number of cases;this,of course,does not affect the quality of the parametrization.The inverse, up to a re-normalization of?rst two components of the result,is given by:

M?1 c (u,v)=

(1?|2u?1|,2u?1,v)if u≥0,

(|2u?1|?1,2u?1,v)if u<0

(7)

Cylinders:impostors(see?g.3)To draw a cylinder impostor we send two pairs of vertices,each located at either end of the axis of the cylinder.After projection the points are displaced in a direc-tion parallel to the image plane and orthogonal to the cylinder axis,to form a rectangular quad.As before,each vertex is assigned a coordi-nate(s,t)∈(±1,±1),which will be interpolated for the fragments.

To lift the burden off the fragment program,the vertex program computes a set of(signed)intermediate values which are constant over all the impostors and will be reused by all the fragments.Some of these values are found in cylinder plane,which is de?ned as the

plane Fig.3.Applying a8×16texture over a cylinder impostor,using a prism mapping M c.For illustration purposes,bilinear interpolation is disabled, except in the last image,and random color values are assigned to texels. Duplicated

Fig.4.Processing for the corners of cylinder impostors.See text. embedding the cylinder axis and orthogonal to the viewing plane(see ?g.4).

In particular,we compute the values of dz,da and(nx,ny)which are,respectively,the offset in the z direction,the offset in v texture position,and the normal,for a fragment in impostor space at s=0,i.e. on the projection of the cylinder axis.For all other fragments,these values must be multiplied by(1?

√

s).The normal,in impostor space, will simply be given by(s,nx,nz).

After adding da,fragments with v lying outside[?1..1]are dis-carded.

Another task performed by the vertex shader is to extend the posi-tion of the projected vertices also along the axis direction to accommo-date for the part of the cylinder extruding from the original impostor space(dotted blue line in?g.4),only for vertices for which da·s is positive.Finally,the vertex shader also computes the offset of the tex-ture position u(the one varying along the diameter of the cylinder). This data consists of an offset angle and is stored and sent to the frag-ment shader as the unit-length complex number(roti,rotr),because that is a space-and computation-ef?cient way to store a2D rotation (we cannot send a single scalar value because of the distortions intro-duced by the mapping M c).

3.4Patch packing and determination of patch size Thanks to the low-frequency nature of the stored signal,we can use a sparse sampling:every visible texel will cover in most cases multi-ple screen pixels.This means that mini?cation?lters are not needed: MIP-mapping is disabled,and so texture patches can be of arbitrary (non power of2)sizes without causing any artifact.We are free to choose the best?tting patches size,according to the number of patches and available texture size.Consequently unused space(at right and bottom borders of the global texture)is usually small.

For simplicity we choose to ignore differences in the sizes of atoms, devoting squared patches the same size s p for each one.If sticks are present,we pack two(optionally three)rectangular stick patches into a s p×s p meta-patch.If k is the total number of patches and meta-patches,and we plan to use a s t×s t texture,we choose s p simply as s t/

√

k .

3.5Accessing texture

Because of the same reason,it is mandatory for us to interpolate be-tween samples.While it would be possible to adopt an ad-hoc texel interpolation schema that accesses several non necessarily adjacent texels,in order to increase performance we prefer to perform a sin-gle,standard bilinearly interpolated texture access per fragment(an

TARINI,ET AL.:AMBIENT OCCLUSION AND EDGE CUEING FOR MOLECULAR VISUALIZA

TION

Fig.5.Rendering a molecule with impostors.Top left:the actual im-postors used are shown as wireframed quads.Cylinders impostors are parallel to the projection of their axis,and ball impostors are screen aligned.Top right:cylindrical impostors have been processed.Bottom:ball impostors have been projected.Intersections between primitives are correctly computed via the zeta-buffer.

heavily optimized operation in graphic cards).If this is done without care,discontinuity of M would became visible as discontinuity arti-facts in the corresponding parts of the primitive (see Fig.6).We solve the problem by offsetting all texture coordinates by half a texel and by replicating some of the texel in each patch:?g.6)shows how the problem is ?xed for M s 2;similar solutions are adopted for M s 1and M c .The amount of texel replication needed by the two sampling is different.With M s 1,a texture patch of size 3n ×2n texels contains (n 3?(n ?2)3)unique texels out of 6?n 2.For M s 2,a n ×n texture has n 2?2n +1unique texels out of n 2.From this point of view M s 2is therefore advantageous.4

GPU C OMPUTATED A MBIENT O CCLUSION

In order to compute ambient occlusion,we ?rst build a set of directions sampling ω.To make it well distributed,we start with a tetrahedron or octahedron and we subdivide it until the desired number of directions is reached,then we apply a Laplacian smoothing to the set of found directions.

Similarly to [27],we perform a pair of off-screen rendering passes for each direction d i in the set.In the ?rst pass a “shadow-map”is produced rendering the z-buffer of molecule using d i as view direction.The next pass is a rendering that writes over the texture for the molecule,and accesses the shadow-map produced in the previous pass.For each primitive we send a quad covering the corresponding texture patch,complete with attributes encoding the position and shape of that primitive.For each produced fragment at position (u ,v ),we compute p =M ?1(u ,v ),then we transform p with the same viewing conditions used during the ?rst pass,and comparing the resulting depth with the one extracted from the shadow-map to determine whether p is lit by light coming from d i .If so,the light contribution for d i ,according to equation 2,is accumulated at the corresponding pixel/texel through alpha blending.

Note that replicated texels (see Sec.6)are dealt with correctly be-cause both copies will be mapped by M ?1in the same 3D position and therefore will produce the same result.

Shadow-map computation is particularly undemanding in our case.It is known that shadow-map,in the case of closed blockers,is more robust if the mid-surfaces (half-way between front-facing and back-facing ones)are drawn.In our case (sphere and cylinders),the mid-surfaces are ?at and can be rendered by simply disabling per-fragment depth displacement during impostor rendering.The fragment pro-grams stops being depth-replacing,and,since shading is disabled as well,reduces to a simple membership test (for spheres)or to a single output operation (for

cylinder).

Fig.6.The continuity problem when using bilinear interpolation to ac-cess texture.In the left column the texture patch is shown in u ?v pa-rameter space.The second and third column show a rendering of the impostor,with the sphere rotated to show the most problematic region,which is its “back”part,where the four corners of the texture patch meet.For illustration purposes,in the middle column we use a closest-sample ?lter;in the right column a standard bilinear-interpolation ?lter is used.Top row:when the entire surface of the texture patch is used,bilinear interpolation shows discontinuity lines across cuts.To solve this prob-lem,the texture patch is shrunk in texture space by half a texel in every direction (middle row),and in all four sides of the patch texels values are mirrored around the middle of the edge (bottom row).Texture is never accessed at positions outside the outlined square.The resulting impos-tor is smooth (bottom right).In this case,a total of 81out of 100texels are unique.

We add another optimization,which works for general shapes,con-sisting in the computation of the contributions of two opposed light directions d i and ?d i at a time.The two corresponding shadow-maps are drawn side to side into a single texture.In the second pass,each 3D point p can be lit by only one of d i and ?d i ,according to normal of p (see equation 2),so a single texture access is performed in any case from the appropriate half of the shadow-map.This effectively halves the processing time for the second passes.4.1

Applications

We found that ambient occlusion adds dramatically to the clarity of the molecular rendering,in all visualization modes.This is true for small molecules (few dozens or few hundreds atoms,?g.8),for medium molecules (few thousands atoms,?g.8),and even more so for larger ones (several tenths of thousands of atoms,?g.7).In the last case,standard direct shading alone,even when enhanced with depth cueing and shadows,can fail to produce an intelligible still image,while the 3D structure becomes evident when ambient occlusion is used,even without any other contribution.5

F URTHER E NHANCIN

G V ISUAL Q UALITY

There are a number of visual effects that are easy to add in our impostor-based framework.5.1

Depth aware contour lines

Since we work in image space,contour lines can be easily added.The ?rst effect consists in drawing a solid line around each primi-tive.Since these lines do not affect the depth value of fragments,they naturally disappear at the intersections of atoms,ad occur only to sep-arate actually detached atoms,boosting the clarity of rendered images (see ?g.9).Less importantly,the border can help to tell the difference between larger vs.just nearer atoms when perspective views are used.

IEEE TRANSACTIONS ON VISUALIZA TION AND COMPUTER GRAPHICS,VOL.12,NO.5,SEPTEMBER/OCTOBER

2006

Fig.7.Example of the ef?cacy of Ambient Occlusion to deliver impres-sion of 3D shape in still images for large molecules.A molecule of 1AON (model taken from [1]),consisting of 58688atoms is shown with (from top left,in Z order):standard direct lighting,direct lighting with depth cueing,direct lighting with cast shadows and depth cueing,and Ambi-ent Occlusion alone.The last image uses a 1024×1024texture with 4×4sized octahedron based patches.

Fig.8.Other comparisons between direct illumination (left)and ambi-ent occlusion (right)for small (above)and medium (below)molecules.Above:testosterone (49atoms)and,below,porin (2219atoms).Models taken from

[1].

Fig.9.Above,left:a rendering with constant width,anti-aliased lines.Above,right:thinker,depth aware lines.Bottom (for comparisons):di-rect illumination,cast shadows,Ambient occlusion.

To achieve solid lines,fragments detected to be just outside the range R of the impostor (namely with a distance d from the center with d between R and R +ξ,for a given parameter ξ)are overwritten with black color rather then discarded.The internal border of the lines can be interpolated between line color and color of the primitive (using (d ?R )/ξas weight)to eliminate pixel aliasing on the internal part of the contour.

For a more informative rendering,the thickness of the lines can be made dependent on the jump in depth between the primitives separated by the contour line (see ?g.9).This effect proved useful in illustrative rendering (in the different context of pen-and-ink illustrations of trees)in [7].

In our approach this can be cheaply done in a single-pass rendering:fragments at a 2D distance d between R and R +ξare pushed back,increasing their depth by the value η(ξ?R ).This produces a truncated cone topped with a semi-sphere,and the depth buffer ensures the desired effect.

To explain why,here we see a 2D “side view”of the depth buffer after that atoms A B C are drawn.Depth-culled frag-ments are shown in green.In the ?nal screen buffer (represented by a line),the black segment separating A and B ap-pears shorter than the one separating B and C,signalling to the viewer that the

depth jump between A and B is smaller.The parameter ηdictates how strongly depth jumps affect thickness (ξ2is the maximal depth jump after which the line thickness stops increasing).Relying solely on the depth buffer,this algorithm is independent of the order of rendering of the primitives.5.2

Halo Effect

The very common way to suggest depth for 2D images (especially use-ful when they represent unfamiliar objects like molecules)is to resort to depth-cueing,where fragments are darkened —or more generally pushed towards a given background color —according to their depth.This,however,has drawbacks:the vision of the furthest parts of an object is hindered by the arti?cial “fog”,which is bound to reduce contrast.Even worst,this effect is distributed over all the images,also far from dif?cult to visualize depth steps.

We propose here to achieve a similar effect,but without unneces-sary losing contrast over the entire image:we draw transparent “halos”

TARINI,ET AL.:AMBIENT OCCLUSION AND EDGE CUEING FOR MOLECULAR VISUALIZA

TION

Fig.10.Halos drawn around molecules to communicate a sense of depth.Left:for illustrative purposes,black halos alone are drawn.Right:a dimly lit rendering combined with white halos.

around each atom.Each point in the halo is more opaque the bigger the distance between it and its background.The halo fades to zero also with the distance from the atom border (see ?g.10).Both darkening and lightening halos can be used.

Not only this helps identifying depth discontinuities,but it also helps when it comes to immediately recognize the general slope of “walls”of atoms (wall slopes cannot be easily identi?ed by direct shading alone because they are composed by primitives).When a wall composed by several atoms is seen from grazing angles,the cumulated effect of their halos communicates so to the viewer.

A similar use of brightening/darkening halos to improve the per-ception of depth discontinuities for general scenes was proposed,in a totally independent way,by Luft et al.[19].

In our approach halos are rendered in a second pass,after the depth buffer is set,and stored in a separate https://www.wendangku.net/doc/ec7244984.html,rger circles are drawn around each atom,as ?at rendering impostors passing through the atom.In this pass we do not affect the zeta of the produced fragments.Only a color is produced according to z -difference and distance from the center.The depth test is enabled but we do not write on the depth buffer.Every rendered fragment darkens (or brightens when white ha-los are used)the color of the corresponding screen pixel,in an order independent way.5.3

Z-clipping of impostors

When the near clipping plane cuts through an atom,the top part of the atom is clipped out of view,leaving visible whatever is behind the atom (background,or other intersecting atoms).The effect harms the clarity of the scene because most people intuitively imagine the atom modeled as “full”solid spheres rather than as thin empty shells.To solve this,when a fragment of an impostor (after z computation)is nearer to the eye than the near clipping plane,we test the depth of the other intersection of the current view ray with the sphere (which is trivially found inverting the z displacement).If also this point falls on the viewer’s side of the clipping plane than the fragment must be discarded.Otherwise,the fragment is moved onto the clipping plane,its normal is overwritten to (0,0,?1),to suggest a ?at surface of a “cut”atom,and the ambient occlusion term is also set to a full lit value.

If the new depths of clipped fragments were all set to the same value,the second of two clipped,intersecting primitives would be drawn over the ?rst.To show a plausible intersection inside the atoms,we set the new depth value of capped atoms to to a small positive value dependent on their original computed depth z .We use ε(K +z )for a small εand a constant K .This way,depth-test correctly computes the intersection (see Fig.11).

We also lighten the ambient occlusion term for fragments close to the clipping plane,to roughly simulate the temporary culling of light-blocker in front of

them.

Fig.11.A detail of a rendering showing a molecule cut by the near clipping plane.

6R ESULTS

We presented a set of rendering techniques that,in our opinion,ef-fectively extends the ability to produce real time molecular renderings which clearly communicate 3D shapes.Some of these techniques,like edge-cueing ones,have not been presented before (to our knowl-edge),whereas the occlusion culling consisted in the adaptation and optimization for the case of impostor-based rendering of a class of methodologies developed for polygonal meshes.These visual effects are not mutually exclusive and are designed to be combined in the same rendering.

The ambient occlusion computation time is always very short,av-eraging 233light directions per second for a medium model of 2219atoms and around 900K effectively used texels,and 15views per sec for the largest we tried of around 60K atoms (on a Athlon -2.6GHz,with an ATI X1600).This means that ambient occlusion can be com-puted interactively if the application requires it.

The proposed impostor based rendering approach proved very ef-fective.In our tests the frame-rate kept in sync with the monitor re-fresh rate in all but the most demanding scenarios (full screen,very large molecules,all effects combined),and it never dropped below 20frames per sec.The main weakness of our approach lies in the depth https://www.wendangku.net/doc/ec7244984.html,e of hierarchical depth buffer structures for quick culling of entire primitives could further accelerate the rendering.One key advancement proposed here is the ability to combine the ?exibility of texture with the ef?ciency of 2D impostors,admittedly only for the special case of spheres and cylinders.This can probably be exploited in other ways as well.

A simple,molecule visualization tool that opens PD

B ?les and de-livers what is described in this paper is publicly available at the project home-page:https://www.wendangku.net/doc/ec7244984.html, .A CKNOWLEDGEMENTS

We wish to thank Enrico Gobbetti and Mario Valle for insightful dis-cussions,on importance of high quality shading in the three dimen-sional perception of complex shapes in space and Monica Zopp`e for having introduced us to the ?eld of Molecular Visualization.R EFERENCES

[1]H.M.Berman,J.Westbrook,Z.Feng,T.N.Gilliland,G.and Bhat,

H.Weissig,I.N.Shindyalov,and P.E.Bourne.The protein data bank.Nucleic Acids Res.,28:235242,2000.https://www.wendangku.net/doc/ec7244984.html,.

[2]R.Borgo,P.Cignoni,and R.Scopigno.An easy to use visualization

system for huge cultural heritage meshes.In D.Arnold,A.Chalmers,and D.Fellner,editors,VAST 2001Conference Proc.,pages 121–130,Athens,Greece,Nov.28-302001.ACM Siggraph.

[3]Michael Bunnell.GPU Gems 2,chapter Dynamic Ambient Occlusion

and Indirect Lighting,pages 223–233.Addison-Wesley,2005.

IEEE TRANSACTIONS ON VISUALIZA TION AND COMPUTER GRAPHICS,VOL.12,NO.5,SEPTEMBER/OCTOBER2006

[4]Tolga Can,Yujun Wang,Yuan-Fang Wang,and Jianwen Su.Fpv:fast

protein visualization using java3d.Bioinformatics,19(8):913–922,2003.

https://www.wendangku.net/doc/ec7244984.html,.

[5]USCF Chimera.https://www.wendangku.net/doc/ec7244984.html,/chimera/.

[6]Cn3D.https://www.wendangku.net/doc/ec7244984.html,/structure/cn3d/cn3d.shtml.

[7]Oliver Deussen and Thomas https://www.wendangku.net/doc/ec7244984.html,puter-generated pen-and-ink

illustration of trees.In SIGGRAPH,pages13–18,2000.

[8]David S.Ebert and Penny Rheingans.V olume illustration:non-

photorealistic rendering of volume models.In IEEE Visualization,pages 195–202,2000.

[9]Bruce Gooch,Peter-Pike J.Sloan,Amy Gooch,Peter Shirley,and

Richard F.Riesenfeld.Interactive technical illustration.In SI3D,pages 31–38,1999.

[10]Andreas Halm,Lars Offen,and Dieter Fellner.Biobrowser a framework

for fast protein visualization.In EUROVIS2005:Eurographics/IEEE VGTC Symposium on Visualization2005,pages.287–294,2005. [11]Andreas Halm,Lars Offen,and Dieter W.Fellner.Visualization of com-

plex molecular ribbon structures at interactive rates.In IV,pages737–744.IEEE Computer Society,2004.

[12]Andrey Iones,Anton Krupkin,Mateu Sbert,and Sergey Zhukov.Fast,

realistic lighting for video games.IEEE Computer Graphics and Appli-cations,23(3):54–64,May/June2003.

[13]Tobias Isenberg,Bert Freudenberg,Nick Halper,Stefan Schlechtweg,

and Thomas Strothotte.A developer’s guide to silhouette algorithms for polygonal models.IEEE Computer Graphics and Applications, 23(4):28–37,2003.

[14]Tobias Isenberg,Maic Masuch,and Thomas Strothotte.3D Illustrative

Effects for Animating Line Drawings.In Proceedings of the IEEE Info-Vis,July19–21,2000,London,England,pages413–418,Los Alamitos, California,2000.IEEE Computer Society.

[15]James T.Kajiya.The rendering https://www.wendangku.net/doc/ec7244984.html,puter Graphics(SIG-

GRAPH),20(4):143–150,1986.

[16]Hayden Landis.Production ready global illumination.In Siggraph2002

Course Notes:,pages331–338,2002.

[17]Michael https://www.wendangku.net/doc/ec7244984.html,nger and Heinrich H.Bulthoff.Perception of shape from

shading on a cloudy day.Technical Report Technical Report No.73, Max-Planck-Institut fur biologische Kybernetik,October1999.

[18] B.K.Lee and Fred M.Richards.The interpretation of protein structures:

estimation of static accessibility.J Mol Biol.,55(3):379–400,Feb1971.

[19]Thomas Luft,Carsten Colditz,and Oliver Deussen.Image enhancement

by unsharp masking the depth buffer.ACM Transactions on Graphics (Siggraph Proc.),25(3),jul2006.to appear.

[20]Lee Markosian,Michael A.Kowalski,Daniel Goldstein,Samuel J.

Trychin,John F.Hughes,and Lubomir D.Bourdev.Real-time nonphoto-realistic rendering.In SIGGRAPH,pages415–420,1997.

[21]Gavin Miller.Ef?cient algorithms for local and global accessibility shad-

ing.In ACM SIGGRAPH’94,pages319–326,New York,NY,USA, 1994.ACM Press.

[22]Matt Pharr.GPU Gems,chapter Ambient occlusion,page667692.

Addison-Wesley,2004.

[23]Bui Tuong Phong.Illumination for computer generated https://www.wendangku.net/doc/ec7244984.html,-

mun.ACM,18(6):311–317,1975.

[24]Emil Praun and Hugues Hoppe.Spherical parametrization and remesh-

ing.ACM Trans.Graph.,22(3):340–349,2003.

[25]Ramesh Raskar and Michael F.Cohen.Image precision silhouette edges.

In SI3D,pages135–140,1999.

[26]Takafumi Saito and Tokiichiro https://www.wendangku.net/doc/ec7244984.html,prehensible rendering of

3-d shapes.In SIGGRAPH,pages197–206,1990.

[27]Mirko Sattler,Ralf Sarlette,Gabriel Zachmann,and Reinhard Klein.

Hardware-accelerated ambient occlusion computation.In VMV,pages 331–338,2004.

[28]R.A.Sayle and https://www.wendangku.net/doc/ec7244984.html,ner-White.Rasmol:biomolecular graph-

ics for all.Trends Biochem.Sci.,20:374376,1995.available at: https://www.wendangku.net/doc/ec7244984.html,/microbio/rasmol/.

[29]Peter-Pike Sloan,Jan Kautz,and John Snyder.Precomputed radiance

transfer for real-time rendering in dynamic,low-frequency lighting envi-ronments.In SIGGRAPH2002:,pages527–536,New York,NY,USA, 2002.ACM Press.

[30] A.James Stewart.Vicinity shading for enhanced perception of volumetric

data.In IEEE Visualization2003(VIS’03),page47,Washington,DC, USA,2003.IEEE Computer Society.

[31]Ivan Viola,https://www.wendangku.net/doc/ec7244984.html,ard Gr¨o ller,Markus Hadwiger,Katja B¨u hler,Bernhard

Preim,Mario Costa Sousa,David S.Ebert,and Don Stredney.Illustrative

visualization.In IEEE Visualization,page124,2005.

[32]Sergej Zhukov,Andrej Inoes,and Grigorij Kronin.An ambient light illu-

mination model.In George Drettakis and Nelson Max,editors,Rendering Techniques’98,Eurographics,pages45–56.Springer-Verlag Wien New York,1998.

立体构成教案 (1)

立体构成 立体构成的概念 除开平面上塑造形象与空间感的图案及绘画艺术外，其它各类造型艺术都应划归立体艺术与立体造型设计的范畴。它们的特点是，以实体占有空间、限定空间、并与空间一同构成新的环境、新的视觉产物。由此，人们给了它们一个最摩登的称谓："空间艺术"。 既然共属于"空间艺术"。那么无论各自的表现形式如何，它们必有共通的规律可循。近年来人们对此进行了不懈的探索，取得了以"立体构成"作为空间艺术基础的经验（类似绘画中的基础是素描、色彩一样），并已由实践证明，它直接有助于创作与设计，是用于基础教学的新学科。了解和研究立体构成，并通过训练，掌握其原理及构成形式、过程和方法，对每一位艺术家或设计家所从事的创作而言，便如同医生熟知了各种药的药性，只待在医疗实践中对症下药了。 一、立体--三次元 （一）、立体的概念 立体是实际占有空间的实体。它较之于在二次元的空间中（即平面中）所表现出来的立体感，是两种截然不同的性质。 平面中表现的空间深度和层次，是单纯视觉的，它运用透视法来表现立体的效果。而立体，则是在空间实际占有位置的实体，我们可以围绕着它变换成任意角度，前后左右地观看。小的立体形态，我们还可以拿在手中翻来覆去地观赏，盲人还可以靠手的触摸体会到它的形象，所以立体的"形"与面的"形"是不能相提并论的。它的形不是绘画平面中的轮廓的概念，而是从不同角度观看时产生的不同形态。 如果根据以前的定义推测，立体应该是面所移动的轨迹。面在移动时，不只是顺着自身长或宽的方向滑动，而是必须朝着和面成角度的方向移动。另外，通过面的旋转也能产生立体。这种动的定义是理念的、概念的，有助于我们对立体形态的理解。 立体的类别及性格 1、立体的类别 立体是有形的实体，而这实体的表面形象均是由线和面组成的（或者说，可以分解成线和面）。如果把各种二次元的形三次元化，就会产生各种相应的立体形，所以，立体形与线、面的形的关系是很密切的。因此与线和面一样，立体也可划分为直线系、中间系、和曲线系三大类。进而立体又可分为几何形立体和自由形立体两大类。综合起来，立体的类别可分为：

win10怎么把edge浏览器设置为默认浏览器方法

win10怎么把edge浏览器设置为默认浏览器方法 方法步骤 1、在win10中我们在开始菜单中找到“设置”打开 2、在打开的设置面板中我们找到“系统” 3、然后在系统下面我们找到左边的“默认应用”，之后你会看到应用对应的“Web 浏览器” 4、点击“选择默认应用”，如下所示我们会看到很多浏览器了，选择Microsoft Edge 设置成默认浏览器

补充：浏览器常见问题分析 1.IE浏览器首次开机响应速度慢，需要数秒。搞定办法：IE下选择工具-internet 选项-连接-局域网设置-取消自动检测。 2. IE9图片显示不正常或干脆不显示，尤其是QQ空间搞定办法：工具-internet 选项-高级-加速图形-运用软件而非GPU 选择。 3. 打开网页显示【Internet Explorer 已不再尝试还原此网站。该网站看上去仍有问题。您可以执行以下操作:转到首页】搞定方案：工具-internet选项-高级中关闭【启用崩溃自动恢复】重新启动ie后即开。 4. 下载完所需安全控件也无法运用各种网银，付款时识别不出u盾搞定方案：据提示下载银行安全控件并安装。插上u盾，拿建行为例：在开始菜单里-所有程序-中国建设银行E路护航网银安全组件-网银盾管理工具打开后点击你的u盾并注册。然后重新启动浏览器(一定要完全退出再进) 进入付款网页上方会显示是否允许加载项，选择在所有站点允许。这时候可能还需要再次重新启动浏览器进入付款页面这时候你期待的u盾密码输入框会出现。这样就ok了 5. 打开网页一直刷新-失败-刷新，无限循环搞定办法：工具-internet选项-高级-禁用脚本调试。 6. IE 习惯性停止工作或崩溃。搞定办法：工具-管理加载项，一一禁用排除以找到某个插件的问题。由于情况多种多样，有些时候找不到具体原因，我们可以通过重置来搞定工具-internet选项-高级。 相关阅读：浏览器实用技巧

怎样修复IE浏览器

1.清理多余的插件 ①安装太多的插件，将直接导致 Internet Explorer（IE）运行效率降低。 特别有些病毒、木马、恶意软件等更是喜欢利用IE插件来做文章。 ②建议使用《360安全卫士》、《瑞星卡卡安全助手》、《金山卫士》等 软件进行查杀、清理、修复。注意：不要只局限使用一款软件，多用几款或许有一款正好适合您。当然最好是在安全模式下进行。这样一般就可以解决病毒问题。 2.超链接无任何反应，IE无法打开，修复受损的IE模块 ①有时候IE无法打开或者IE无法打开新窗口了，表现为：在浏览网页过 程中，单击超链接无任何反应。问题的原因在于IE新窗口模块被损坏所致，解决的方法是单击“开始”－“运行”,依次运行： regsvr32 actxprxy.dll regsvr32 shdocvw.dll ②将上述两个DLL文件注册，然后重启系统。如果还不行，则需要继续执 行： regsvr32 mshtml.dll regsvr32 urlmon.dll regsvr32 msjava.dll regsvr32 browseui.dll regsvr32 oleaut32.dll regsvr32 shell32.dll 3.清理使用痕迹 ①方法1，软件。下载并安装《360安全卫士》或者《金山卫士》等软件， 然后使用其“清理使用痕迹”功能进行清理。 ②方法2，手工。“开始－运行”，输入 regedit，回车。 在弹出的注册表编辑窗口，单击左边的“我的电脑”，然后按键盘上的“F3”键。输入浏览记录里面的一条（注意文字要完全一致），然后点击“查找”。 找到后，对其点右键，选择“删除”。估计同类的记录都在一起，一并删除之。 然后继续按“F3”，直到找不到为止。 4.清除IE缓存 ①打开IE，在“工具”菜单上，单击“Internet 选项”。 ②在弹出窗口中，单击“常规”选项卡，在“Internet临时文件”下。 单击“删除文件”。 ③然后在弹出窗口中，钩选“删除所有脱机文件”。单击“确定”。 ④单击“删除Cookies”。单击“确定”。 ⑤在“历史记录”下，单击“清除历史记录”，然后单击“是”。单击“确 定”。

如何维护老客户与开拓新增客户

如何维护老客户与开拓新增客户 导读： 准客户开发只有两种途径，一是自己开发，一是客户帮你开发，所以请你每天去回访你的客户。 “建立影响力中心，充分运用转介绍，促使你的寿险事业如日中天。因为： 第一，推荐你的同事或街坊拥有保险保障，你们单位你们小区就能成为保险公司的"大户"； 第二，帮助你的亲朋好友拥有保险的利益，其实就是帮助你自己，因为他们的问题同样会成为你的困扰。如果他不向保险公司投保，就等于向你投保，而且完全免费！请问你愿意这种情况发生吗？” 保险服务包括售前服务、售中服务和售后服务，也就是说，客户服务工作贯穿我们从事保险工作的始终。如何才能做到、做好客户服务工作？我们必须首先从心态和行为上做——保险生活化，生活保险化。 显然，售后服务是维护老客户的必备方式和有效手段，维护好老客户其实正是增添新客户的高效前提！售后服务的另一种说法叫做“客户回访”，准客户开发只有两种途径，一是自己开发，一是客户帮你开发，所以请你每天去回访你的客户。

一个成功的代理人每天都会拨出专门的时间去拜访他的客户，让客户帮助自己开发准客户，所以他会越做越轻松，越做越有成就感。一个失败的代理人每天都会拨出全部的时间去拜访他的准客户，只能依赖自己开发准客户，所以他会越做越辛苦，越做越有挫折感。 做好客户回访的关键其实并非在如何回访？因为人的不同，方式方法亦不一而同，关键在于“去”回访，在于行动，在于走出去，走进客户。保险营销员在其职业生涯当中,常常会出现三大致命伤,首先是”懒”,这其实就是失败的开始；其次是“傲”，这是保险业务员职业生涯中的第一大病，其实，每个代理人，在发展的过程中，要学会谦逊，在获得成绩时，掌声愈大、腰要愈软；三是“伪”，这是人际关系的致命伤。 作为一个成功的寿险营销员，要做到“人生三勤”。即勤能补拙、勤能补运、勤能补情。 勤能补拙。笨鸟先飞，成事靠准备，不是靠智慧。勤能补运（时）。客户被拜访99次，第100次被你访到，幸运成交。勤能补情。多与顾客接触，能增近感情。 加强客户回访，要成为我们日常工作的好习惯。成功的人都有好习惯，好习惯是勉强来的，勉强成习惯，习惯成自然。因此，加强客户回访，不断给客户以保险的观念与理财分析，你的诚心，必将能打动客户。我们要让客户明白，买保险不仅要买单一的保险产品，更要买组合型的产品，买全方位的保障。我们要让客户明白，不仅自己要买保险，自己身边的关系人也必须要购买保险。保险应当成为人们生活当中的必须品，人人都该拥有足够的保险保障，从而获得自立、自尊的尊贵生活。保险从业人员应充分运用保险的这一独特功用，为更多的客户提供保险保障服务，从而使我们自己的寿险从业生涯走向良性循环。

浏览器兼容性问题及解决方案

浏览器兼容新问题 W3C对标准的推进，Firefox，Chrome，Safari，Opera的出现，结束了IE雄霸天下的日子。 然而，这对开发者来说，是好事，也是坏事。 说它是好事，是因为浏览器厂商为了取得更多的市场份额，会促使各浏览器更符合W3C标准，而得到更好的兼容性，并且，不同浏览器的扩展功能(例如-moz，-webkit开头的样式)，对W3C标准也是个推进；说它是坏事，因为，多个浏览器同时存在，这些浏览器在处理一个相同的页面时，表现有时会有差异。这种差异可能很小，甚至不会被注意到；也可能很大，甚至造成在某个浏览器下无法正常浏览。我们把引起这些差异的问题统称为“浏览器兼容性问题”。而正是这些“浏览器兼容性问题”，无形中给我们的开发增加了不少难度。 从浏览器内核的角度来看，浏览器兼容性问题可分为以下三类： 1. 渲染相关：和样式相关的问题，即体现在布局效果上的问题。 2. 脚本相关：和脚本相关的问题，包括JavaScript和DOM、BOM方面的问题。对于某些浏览器的功能方面的特性，也属于这一类。 3. 其他类别：除以上两类问题外的功能性问题，一般是浏览器自身提供的功能，在内核层之上的。 例如下面的页面，是一个渲染相关的问题： 在各个浏览器中都表现的不同，这就属于兼容性问题。 造成浏览器兼容性问题的根本原因就是浏览器各浏览器使用了不同的内核，并且它们处理同一件事情的时候思路不同。 现今常见的浏览器及其排版引擎（又称渲染引擎）及脚本引擎，如下：

而造成浏览器兼容性问题的常见原因则是设计师写出了不规范的代码，不规范的代码会使不兼容现象更加突出。 例如： 不规则的嵌套： DIV 中直接嵌套LI 元素是不合标准的，LI 应该处于UL 内。此类问题常见的还有P 中嵌套DIV，TABLE等元素。 不规范的DOM接口和属性设置： 总之，人为的原因也占很大一部分。而人为造成兼容性问题的原因，除了粗心之外，大都源于浏览器bug 的存在，和开发者对标准的不了解。 比如，如果要做一个功能，功能是想让鼠标悬停在IMG 元素上方时，可以出现提示信息，经常针对IE 做开发的人，可能会使用IMG 元素的“alt” 属性，但其他浏览器中就是不给…alt? 属性面子。因为W3C 标准中规定要去做这件事的属性是”title“，大多浏览器符合标准，IE 不符合，这是IE 浏览器内核的问题；开发者不知道”title“ 才是正解，不遵循标准去写代码，是开发者的问题。所以，一个问题分两半，浏览器和开发者都有责任。 既然都有责任，就都有义务去解决兼容性问题。那么，从浏览器的角度来讲，它的厂商应该修复浏览器的bug和不合标准的地方，当某一天IE 的”alt“ 不能用于提示了，还有人用这个错误的属性去显示提示么？从开发者角度来讲，多了解标准，了解浏览器兼容性问题，就可以在开发的过程中，有效的避开兼容性问题，让你的页面在所有浏览器中畅通无阻。 废话少说，下面就讲讲如何有效的避免一些兼容性问题。 J AVA S CRIPT (4)

客户的维护与开发

读《怎样做好客户保持》后感 -------胡玉彬 客户是一个公司的根本，也是企业赖以发展的源泉，一个企业的持续发展，除了正常的管理，经营；客户的开发和维护显然是重中之重。 开发一个客户和维护一个客户，大家都知道，从经济学角度来讲，很明显，维护客户能够大大的节约成本，正如文中所讲“开发一个新客户所需要花费的成本是维护一个老客户的5-10倍，客户保持比吸引新客户更能够降低成本”。 我们只需要花开发新客户1/3的精力，1/5的成本，不仅能维护好客户，还能在现有客户的基础上越做越大，因为，客户也要发展。一个一点不图发展的客户，不是一个好客户。我们要在有限的资源和精力情况下，最大效益的保持最有价值的客户。 根据已经成交的客户使用情况，整理归纳，有一个清晰明了的客户信息数据库，知道哪些是现有最大客户利益化，哪些是最有潜力，哪些是长期维护的客户，根据不同客户的情况，有自己的一套维护体系，有限的资源，有效的时间，效果最好。 中国的社会，归根到底，是人情的社会，所以要重视与客户情感的建立，正如文中所言“把客户感情的维护与企业提供的产品和服务紧密的联系在一起”，这样才能大大提高客户对公司，对产品的认可度和依赖度。 客户的维护，没有客户开发那样有难度，有刺激，和成就感，相

比而言，维护客户相对比较平淡和持久。这就更要求我们自己要注意细节，注意自己内心对客户的理解和尊敬，客户是自己的衣食父母，只有在日久的维护中，始终能保持一颗关心，帮助客户的心态，才能真正让客户感觉到，我们是在和他们共成长，才会有牢靠的客情关系。 用真心和耐心去维护客户，用信心和激情去开发客户，在维护中学习，从开发中反思，做出属于自己的品牌营销！

快速修复浏览器方案

快速修复浏览器方案 (鉴于系统环境不同→请活学活用以下方法→根据具体情况决定做哪些→并非都做到) 1、打开浏览器，点“工具”→“管理加载项”那里禁用所有可疑插件,或者你能准确知道没问题的保留。然后→工具→INTERNET选项→常规页面→删除cookies→删除文件→钩选删除所有脱机内容→确定→设置使用的磁盘空间为：8MB或以下(我自己使用1MB)→确定→清除历史纪录→网页保存在历史记录中的天数：3以下→应用确定（我自己使用的设置是0天）。 2、还原浏览器高级设置默认值：工具→INTERNET选项→高级→还原默认设置。 3、恢复默认浏览器的方法“工具”→Internet选项→程序→最下面有个“检查Internet Explorer是否为默认的浏览器”把前面的钩选上,确定。 4、设置主页：“工具”→Internet选项→常规→可以更改主页地址→键入你喜欢的常用网址→应用。 5、如果浏览器中毒就使用卡卡助手4.0版本修复,然后做插件免疫：全部钩选→免疫。然后→全部去掉钩选→找到“必备”一项，把能用到的插件重新钩选→取消免疫。能用到的就是FLASH和几种播放器的，其余的不要取消免疫。完成所有操作以后，你的浏览器就不会出问题了。 6、运行→regedit→进入注册表, 在→ HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer \ShellExecuteHooks 这个位置有一个正常的键值{AEB6717E-7E19-11d0-97EE-00C04FD91972}, 将其他的删除（默认项也保留无法删除）。 7、检查你的浏览器是否被某种（游戏或其它）安装程序恶意附加了某种插件→卸载清理掉它。 8、HOSTS文件被修改常常会导致类似问题：文件位置 C:\WINDOWS\system32\drivers\etc 把hosts用记事本打开，内容清空，只保留127.0.0.1 localhost 这一个条目关闭保存。 9、(情况较严重，上面方法不能解决时使用)开始→运行→CMD→窗口内粘贴如下命令:

WIN10系统EDGE浏览器隐藏设置的开启方法

Win10系统的一个创新就是带来了全新的革命性产品：Edge浏览器，那么，已经将电脑系统升级为Win10系统的网友对Edge浏览器的使用还顺手吗?今天小编就跟大家分享Edge 浏览器的一些隐藏设置，这些隐藏设置在普通的设置页面中是找不到的，想要使用这些隐藏设置就要用一些特殊的方法。下面就拉教大家如何使用Win10系统Edge浏览器下的隐藏设置? 比如有时会遇到一些在设置里也无法改变的问题，比如在浏览一些 Edge 明明兼容的网站时，仍被提示使用 IE11 打开;找不到 JavaScript 功能开启入口等等，其实这些设置都被隐藏了。 1、开启Microsoft Edge的开发者设置和实验性功能：开启Microsoft Edge ，在地址栏输入about:flags，回车，如图： 2、关闭Microsoft Edge使用Internet Explorer打开网站的提示，如果在使用edge 的过程中，不想收到使用 IE 打开这个网站的提示，就可以将使用 Microsoft 兼容性列表取消取中即可，这样如下图所示的提示就不会看到了。如果经常使用网银，建议保留此选项，如图： 3、开启asm.js提高执行速度：一些依赖于JavaScript 的网站或者网络游戏，在开启asm.js 之后，会使效率提高。因此，可以在实验性功能中找到 JavaScript ，将其包含的两个选项选中，重启浏览器即可生效，如图： 4、其它与重置：其余关于回环/辅助功能/触摸功能以及 CSS 开发选项，可以根据自己的需要点选。如果忘记了初始设置，没关系，还有页面最上的将所有标志重置为默认。 以上就是如何使用Win10 Edge浏览器下的隐藏设置的介绍了，使用这些隐藏设置的功能，能让我们队Edge这款浏览器有个更全面深刻的认识。

浏览器异常的解决部分办法

浏览器异常的解决部分办法 这里有很多Regsvr32命令,看了后就明白了Regsvr32命令修复系统故障实例使用过activex的人都知道，activex不注册是不能够被系统识别和使用的，一般安装程序都会自动地把它所使用的activex控件注册，但如果你拿到的一个控件需要手动注册怎么办呢？如果修改注册表那就太麻烦了，在windows的system文件夹下有一个regsvr32.exe的程序，它就是windows自带的activex 注册和反注册工具。 2000系统的regsvr32.exe在winnt\system32文件夹下； WInXP系统的regsvr32.exe在windows\system32文件夹下 regsvr32的用法为： "regsvr32 [/s] [/n] [/i(:cmdline)] dllname”。其中dllname为activex控件文件名，建议在安装前拷贝到system文件夹下。 参数有如下意义： /u——反注册控件 /s——不管注册成功与否，均不显示提示框 /c——控制台输出 /i——跳过控件的选项进行安装(与注册不同) /n——不注册控件，此选项必须与/i选项一起使用 执行该命令的方法： 1、可以在“开始”--“运行”，调出运行的对话框，也可以使用Win+R热键，然后直接在输入栏输入即可 2、在开始--运行输入cmd，调出…命令提示符?窗口，然后再执行regsvr32命令。 二、Regsvr32错误消息的说明 当使用Regsvr32.exe 时，它会尝试加载该组件并调用它的DLLSelfRegister 函数。如果此尝试成功，Regsvr32.exe 会显示一个指示成功的对话框。如果此尝试失败，Regsvr32.exe 会返回一条错误消息，其中可能会包括一个Win32 错误代码。 以下列表介绍了RegSvr32 错误消息和可能的原因。 Unrecognized flag:/invalid_flag 键入的标志或开关组合无效（请参阅本文中的“Regsvr32.exe 的用法”一节）。No DLL name specified.

最新整理win10怎么把edge浏览器设置为默认浏览器方法

w i n10怎么把e d g e浏览器设置为默认浏览器方法使用w i n10的用户都知道e d g e浏览器，使用起来很方便，界面也友好，那么怎么设置为我们电脑的默认浏览器呢?下面，小编就为大家介绍下w i n10e d g e浏览器设置为默认浏览器方法。 方法步骤 1、在w i n10中我们在开始菜单中找到设置打开 2、在打开的设置面板中我们找到系统 3、然后在系统下面我们找到左边的默认应用，之后你会看到应用对应的W e b浏览器 4、点击选择默认应用，如下所示我们会看到很多浏览器了，选择M i c r o s o f t E d g e设置成默认浏览器补充：浏览器常见问题分析 1.I E浏览器首次开机响应速度慢，需要数秒。搞定办法：I E下选择工具-i n t e r n e t选项-连接-局域网设置-取消自动检测。 2.I E9图片显示不正常或干脆不显示，尤其是Q Q 空间搞定办法：工具-i n t e r n e t选项-高级-加速图形-运用软件而非G P U选择。 3.打开网页显示搞定方案：工具-i n t e r n e t选项-

高级中关闭重新启动i e后即开。 4.下载完所需安全控件也无法运用各种网银，付款时识别不出u盾搞定方案：据提示下载银行安全控件并安装。插上u盾，拿建行为例：在开始菜单里-所有程序-中国建设银行E路护航网银安全组件-网银盾管理工具打开后点击你的u盾并注册。然后重新启动浏览器(一定要完全退出再进)进入付款网页上方会显示是否允许加载项，选择在所有站点允许。这时候可能还需要再次重新启动浏览器进入付款页面这时候你期待的u盾密码输入框会出现。这样就o k了 5.打开网页一直刷新-失败-刷新，无限循环搞定办法：工具-i n t e r n e t选项-高级-禁用脚本调试。 6.I E习惯性停止工作或崩溃。搞定办法：工具-管理加载项，一一禁用排除以找到某个插件的问题。由于情况多种多样，有些时候找不到具体原因，我们可以通过重置来搞定工具-i n t e r n e t选项-高级。 相关阅读：浏览器实用技巧 现在打开了台式电脑桌面上的360安全浏览器的主页。点击360安全浏览器顶部菜单,可以看到一个剪刀形状的功能扩展的三角形的下拉菜单,在下拉菜单中显

立体构成讲义

立体构成讲义 第一章立体构成概述（2节课） 一、教学目的和任务 立体构成是在三维世界中创造具有深度、广度和体量的形态及空间。通过教学使学生了解立体构成在于把艺术或设计上所需要的各种形态巧妙地配置在指定的空间中，以材料和纯粹的抽象形态为素材，按照视觉效果，运用力学的原理进行组合，促使艺术构想源源不断，并使自己的感觉获得提升。 立体构成是现代设计领域中一门基础造型课，也是一门艺术创作设计课。在立体造型中首先需要明确一个概念，即形态与形状的区别，平面造型中我们称平面的是形状，这个形状是物象的外轮廓。在立体造型中形状是指立体物在某一距离、角度、环境条件下所呈现的外貌，而形态是指立体物的整个外貌。即形状是形态的诸多面向中的一个面向，形态则是诸多形状构成的统和体。形态是立体造型全方位的印象，是形与神的统一。 作为研究形态创造与造型设计的独立学科。所涉及的学科建筑设计、室内设计、工业造型、雕塑、广告等设计行业。除在平面上塑造形象与空间感的图案及绘画艺术外，其它各类造型艺术都应划归立体艺术与立体造型设计的范畴。它们的特点是，以实体占有空间、限定空间、并与空间一同构成新的环境、新的视觉产物。由此，人们给了它们一个最摩登的称谓："空间艺术"。 (PPT1)立体构成也称为空间构成。立体构成是以一定的材料、以视觉为基础，以力学为依据，将造型要素，按照一定的构成原则，组合成美好的形体。它是研究立体造型各元素的构成法则。其任务是，揭开立体造型的基本规律．阐明立体设计的基本原理。 立体构成是由二维平面形象进入三维立体空间的构成表现，两者既有联系又有区别。联系的是：它们都是一种艺术训练，引导了解造型观念，训练抽象构成能力，培养审美观，接受严格的纪律训练；区别的是：立体构成是三维度的实体形态与空间形态的构成。结构上要符合力学的要求，材料也影响和丰富形式语言的表达。立体是用厚度来塑造形态、它是制作出来的。同时立体构成离不开材料、工艺、力学、美学，是艺术与科学相结合的体现。 3、立体构成与电脑辅助设计： 设计是包括立体构成在内，并考虑其它众多要素使之成为完整造型的活动，设计的领域非常广泛，它可分为商业设计、工业设计、环境设计等。而这些艺术门类又可细分为广告设计、书籍设计、包装设计、展示设计、服装设计、染织设计、室内外设计、动漫设计等专业门类。以上的专业无一例外，现在都把计算机技术作为工具设计的重要手段，这无疑具有划时代的意义。 Auto CAD 是较通用的图形设计软件，主要用于二维三维坐标空间，能精确地设计各种几何形体， 3D MAX.是应用最广泛的三维动画软件，并配有丰富的材质库与各种光源，环境效果，设计者可以设计出实体动态影像效果。Maya是经典的三维动画软件，它为三维动画领域提供了一整套特效解决方案，可以逼真的模拟现实生活中的场景。它标志着三维软件又迈入了一个崭新的阶段

彻底修复 IE 的自动完成

彻底修复 IE 的“自动完成”功能 一、故障现象： 我习惯用XP，而且是Ghost版的，很好用！但是，有时候系统的IE中“自动完成”功能不能使用，任凭我怎么折腾都不出来，下面先介绍我具体的折腾方法，当然你也可以试试，如果你的系统病的不重的话用这两着就差不多好了，也许！ 二、初步处理： 1.首先确认：［工具］->［Internet选项］->［内容］->［自动完成］->将四个选项都打勾，并且把下面的“清除表单”和“清楚密码”都点一击，然后点完成。 2.再次确认：开始运行输入gpedit.msc打开组策略，找到用户配置－管理模板－Windows 组件－Internet Exporer，将“禁用表单的自动完成功能” 这一项改成未配置。 组策略打开时如下图：

请注意：有的系统中组策略编辑文件被去掉了，特别是一些Ghost版的系统中会有这样做法。 三、彻底解决： 1. 运行注册表编辑器regedit 2. 找到下面的注册表项：HKEY_CURRENT_USER\Software\Microsoft\Protected Storage System Provider 3. 在右键菜单上，单击权限。 4. 单击高级按钮，. 选中“从父项继承那些......”和“用在此显示的......”复选框。 5. 单击确定，然后在提示您是否继续操作时单击是。 6. 点击展开Protected Storage System Provider 项，单击其下面相关的子项文件夹，在子项文件夹上右击后选择“删除”，然后在警告消息对话框中单击是。 注意： a. 你看到的子项文件夹和这个形式相似： S-1-5-21-124525095-708259637-1543119021-16701，可能略有不同，属正常现象。 b. 对于您的每一个身份标识(多人多个帐户用一台机器)，在Protected Storage System Provider 项下都可能有一个子项。若要在所有身份标识中解决此问题，您必须删除Protected Storage System Provider 项下的所有用户子项文件夹。 7、退出后重启一下电脑，问题解决。

Win 7 IE浏览器修复

在很多时候，用户使用IE上网时，IE浏览器会出现短暂性的假死，如果严重一点的话，只能强制关闭IE浏览器。对于经常计算机裸奔的用户，很可能IE浏览器就受到了有害木马程序的劫持。目前甚至在Windows 7系统中也不会例外。 为了解决这些有害程序的侵入而导致的IE浏览器的故障，我们总结了两个长驱直入的方法，能直捣黄龙将有害侵害剿灭。 处理方式一：如果您的系统是Windows vista\Windows 7 修复方法 先Windows 7功能键win+r，打开运行程序，并输入:regedit 运行regedit 找到: 计算机\HKEY_CURRENT_USER\SOFTWARE\MICROSOFT\WINDOWS\currentversion\Explorer\Fil eExts\.lnk

regedit 你会发现有openwithlist 和openwithprogids 两项，如果有其他的选项将其删除（如上图所示） 再将openwithlist内的除默认以外的所有键值都删除。将openwithprogids内的除默认和lnkfile以外的所有键值都删除。保存退出即可。 处理方式二：打开注册表，找到HKEY_USERS\S-1-5-21-3018251054-2002470566-3220277462-1000\Software\Microsoft\Wind ows\CurrentV ersion\Explorer\FileExts\.lnk\UserChoice

新建项 右侧的Progid键值Applications\\notepad.exe正是记事本，右键将其删除却提示错误不能删除，于是只得将UserChoice这项删除，再右键点击 HKEY_USERS\S-1-5-21-3018251054-2002470566-3220277462-1000\Software\Microsoft\Wind ows\CurrentV ersion\Explorer\FileExts\.lnk

Edge浏览器的28个快捷键

Edge浏览器的28个快捷键： Ctrl + D：将当前页面添加到收藏夹或阅读列表 Ctrl + E：在地址栏中执行搜索查询 Ctrl + F：在页面上查找 Ctrl + H：打开历史记录面板 Ctrl + G：打开阅读列表面板 Ctrl + I：打开收藏夹列表面板(测试好像不起作用) Ctrl + J ：打开下载列表页面 Ctrl + K：重复打开当前标签页 Ctrl + L或F4或Alt + D：选中地址栏内容 Ctrl + N：新建窗口 Ctrl + P：打印当前页面 Ctrl + R或F5：刷新当前页 Ctrl + T ：新建标签页 Ctrl + W：关闭当前标签页 Ctrl + Shift + P：新建InPrivate(隐私)浏览窗口 Ctrl + Shift + R：进入阅读模式(需要网页提供支持) Ctrl +1, 2, 3,…, 8：切换到指定序号的标签 Ctrl + 9：切换到最后一个标签 Ctrl +加号(+)：页面缩放比例增加25% Ctrl +减号(-)：页面缩放比例减小25% Ctrl + 0 ：重置页面缩放级别，恢复100%

Backspace回退键或 Alt +方向左：回退到上一页面 Alt +方向右：前进到下一页面 Esc：停止加载页面 Ctrl +鼠标左键单击：在新标签中打开链接 Ctrl + T ab ：切换到下一个标签 Ctrl + Shift + T ab：切换到上一个标签 Ctrl + Shift +鼠标左键单击：在新标签页中打开链接，并导航至新标签页(测试窗打开后页面未加载) Alt + Shift + click：在新窗口中打开链接

立体构成教案 详案

《立体构成》教案（2017年-2018年第一学年度）系部：应用技术系 编制人：姜静

章节名称绪论 授课方法 和手段 课堂讲授与视频赏析教学 教学目的与要求目的和要求： 1.了解构成教育的重要性 2.了解形态与形式的区别 3.了解构成的含义 4.熟悉构成的源流 5.形态构成的基础 6.形态构成教育的范围 教学基本内容 纲要了解构成教育的重要性 1)从时代发展分析现代设计的百年变迁 2) 3)社会发展带来的新行业——现代设计行业 4) 5)构成教育所需的科学领域 了解形态与形式的区别 通过问答形式分析形态与形式的不同 问：形状与形态的区别？形态和形状是否都具有立体感？ 答：形状是平面的，形态是立体的。 它们都具有立体感，形状的立体感是通过透视原理创造的虚幻空间或矛盾空间。形态是通过自身的运动变化或观者的位置变化所形成的空间体验。

了解构成的含义 构成是只一定材料的形态元素，按照视觉规律、力学原理、心理特性、审美法则进行的创造性的组合。 熟悉构成的源流 1)20世纪初苏联的构成主义运动. 2)包豪斯学校（德国1919~1933） 3)包豪斯的创造者是沃尔特·格罗皮乌斯 4)包豪斯对现代设计教育的贡献 （１）艺术与技术结合 （２）在设计中提倡自由创造，反对模仿抄袭、墨守成规 （３）强调实际动手能力与理论素养并重，开始双轨制教学模式（形象大师、作坊大师）（1925年包豪斯从魏玛迁校到德绍） （４）将学校教育与社会实践结合起来 （５）创造基础构成教学模式（三大构成是包豪斯对现代设计教育模式最大的贡献之一） 形态构成的基础 1)首先是分解的过程，即将复杂的视觉表象彻底分解还原成为单纯的 造型元素（点、线、面、体、空间等造型元素） 2)整合的过程（依据一定的形式法则将造型元素整合为符合视觉传达 目的的形态。） 形态构成教育的范围 1)拓展思维空间，培养空间思维能力 2)培养三维空间的造型能力 3)提高构思创意能力 4)提高对材料和工艺的理解和思考 5)增强造型审美形式的感受能力 欣赏：《工业设计》影片 教学重点与难点重点：形态与形式的区别；难点：形态构成的基础。

恢复桌面ie浏览器图标方案

恢复桌面ie浏览器图标方案 方法一：控制面板->Internet选项->高级->勾上“在桌面显示Internet Explorer”->确定 这是很早的一种方法，适用于IE5.5及之前的版本，而现在大部分用户都在使用IE6.0及更高的版本了，所以现在这个方法基本没什么用处。 方法二：桌面点击右键->属性->桌面->自定义桌面->勾上“Internet Explorer”->确定 这是一个很普遍也很常用的方法，适用于大多数人现在使用的环境，WindowsSP2，IE6.0的情况，不过当用户升级到IE7.0后，原本那个“Internet Explorer”勾选框就给空了出来，被去掉了，不能用鼠标进行选择了。 方法三：桌面点击右键->属性->桌面->自定义桌面->在键盘上按一次“i”键(或者ALT+I)->确定 这个方法适用于除Vista之外所有情况。此方法很巧妙，归避了因为升级到IE7.0而消失的“Internet Explorer”勾选框，其实那个去掉的“Internet Explorer”勾选框仍然在那里，只是用户看不见而已，微软仍然保留了焦点，按“i”的作用就是让焦点聚到“Internet Explorer”勾选框上(虽然看不见)，作用相当于用鼠标点击“Internet Explorer”勾选框，效果一样。 方法四：在任意一个文件夹里新建一个文件夹，然后把这个新文件夹重命名为“Internet Explorer.{871C5380-42A0-1069-A2EA-08002B30309D}”(引

号内的部分)，最后把命名后的文件夹拖到桌面，IE图标就会重新显示出来了。 这个方法相对麻烦一些，也不利于初级电脑用户操作，所以不推荐采用。 方法五：右键点击我的电脑->资源管理器->在窗口左侧选择“桌面”->把这里的IE图标拖到桌面上即可。 这是一个很不错的方法，操作也很简便，适用于除Vista之外任何版本的Windows操作系统和任何版本的IE浏览器。虽然这里的桌面内容和真正的桌面内容是一样的，不过资源管理器里桌面上的系统图标是删不掉的，比如“我的文档”、“我的电脑”，当你删除之后，刷新一下，图标即刻又会出现。所以我认为这种方法是最可靠的，建议大家采用。 方法六：注册表创建 经过不断的尝试和查找资料，在DOS下找回IE图标，覆盖shell32.dll找回IE图标...最终，找到了正确的方法，请按以下方法进行，一定可以： 打开记事本，复制以下内容： 以下内容为程序代码 Windows Registry Editor Version 5.00 [HKEY_CLASSES_ROOT\CLSID\{B416D21B-3B22-B6D4-BBD3-BBD452DB3D5B}] @="Internet Explorer" "InfoTip"="@C:\\WINDOWS\\system32\\zh-CN\\ieframe.dll.mui,-881" "LocalizedString"="@C:\\WINDOWS\\system32\\zh-CN\\ieframe.dll.mui,-880"

(完整版)立体构成教案—详案

《立体构成》 教案 2017 年-2018 年第一学年度） 系部：应用技术系 编制人：姜静

章节名称绪论 授课方法和手 段 课堂讲授与视频赏析教学 教学目的与要 求目的和要求： 1. 了解构成教育的重要性 2. 了解形态与形式的区别 3. 了解构成的含义 4. 熟悉构成的源流 5. 形态构成的基础 6. 形态构成教育的范围 了解构成教育的重要性 3)构成教育所需的科学领域 了解形态与形式的区别 通过问答形式分析形态与形式的不同 问：形状与形态的区别？形态和形状是否都具有立体感？答：形状是平面的，形态是立体的。 它们都具有立体感，形状的立体感是通过透视原理创造的虚幻空间或矛盾空间。形态是通过自身的运动变化或观者的位置变化所形成 教学基本内 容 纲要1) 从时代发展分析现代设计的百年 变迁 2) 社会发展带来的新行业——现代设计行业

的空间体验。 了解构成的含义构成是只一定材料的形态元素，按照视觉规律、力学原理、心理特性、审美法则进行的创造性的组合。 熟悉构成的源流 1）20 世纪初苏联的构成主义运动. 2）包豪斯学校（德国1919~1933） 3）包豪斯的创造者是沃尔特·格罗皮乌斯 4）包豪斯对现代设计教育的贡献（１）艺术与技术结合（２）在设计中提倡自由创造，反对模仿抄袭、墨守成规（３）强调实际动手能力与理论素养并重，开始双轨制教学模式（形象大师、作坊大师）（1925 年包豪斯从魏玛迁校到德绍）（４）将学校教育与社会实践结合起来（５）创造基础构成教学模式（三大构成是包豪斯对现代设计教育模式最大的贡献之一） 形态构成的基础 1）首先是分解的过程，即将复杂的视觉表象彻底分解还原成为单纯的造型元素（点、线、面、体、空间等造型元素）2）整合的过程（依据一定的形式法则将造型元素整合为符合视觉传达目的的形态。） 形态构成教育的范围 1）拓展思维空间，培养空间思维能力

立体构成概念及起源

立体构成概念及起源 概念： 立体构成是现代设计领域中一门基础造型课，也是一门艺术创作设计课。在立体造型中首先需要明确一个概念，即形态与形状的区别（知识点一）平面造型中我们称平面的行为形状，这个形状是物象的外轮廓。在立体造型中形状是指立体物在某一距离、角度、环境条件下所呈现的外貌，而形态是指立体物的整个外貌。即形状是形态的诸多面向中的一个面向，形态则是诸多形状构成的统和体。形态是立体造型全方位的印象，是形与神的统一。 作为研究形态创造与造型设计的独立学科。所涉及的学科建筑设计、室内设计、工业造型、雕塑、广告等设计行业。除在平面上塑造形象与空间感的图案及绘画艺术外，其它各类造型艺术都应划归立体艺术与立体造型设计的范畴。它们的特点是，以实体占有空间、限定空间、并与空间一同构成新的环境、新的视觉产物。由此，人们给了它们一个最摩登的称谓："空间艺术"。 立体构成也称为空间构成。立体构成是以一定的材料、以视觉为基础，以力学为依据，将造型要素，按照一定的构成原则，组合成美好的形体。它是研究立体造型各元素的构成法则。其任务是，揭开立体造型的基本规律．阐明立体设计的基本原理。 立体构成是由二维平面形象进入三维立体空间的构成表现，两者既有联系又有区别。联系的是：它们都是一种艺术训练，引导了解造型观念，训练抽象构成能力，培养审美观，接受严格的纪律训练；区别的是：立体构成是三维度的实体形态与空间形态的构成。结构上要符合力学的要求，材料也影响和丰富形式语言的表达。立体是用厚度来塑造形态、它是制作出来的。同时立体构成离不开材料、工艺、力学、美学，是艺术与科学相结合的体现。 “立体构成” 这门课程起源于 1919 年，是德国包豪斯学院在创办后确立的艺术流派： 包豪象斯构成理论: 在现代设计史上，包豪斯构成理论及其教育体系具有特殊的时代意义。成立于 1919 年的包豪斯设计学院，不仅有在建筑史上被称为设计经典的校舍，更有独树一帜的设计理念和教育思想。它奠定了现代工业设计的基础、成为现代设计师的摇篮．以至有人评价它是现代设计真正的开端。 包豪斯构成理论的产生是社会发展的必然，欧洲的产业革命为它的产生奠定了强大的物质基础。社会变革是新思想、新观念的催生婆，英国的产业革命在由手工转向机械化生产的过程中，由于传统观念的影响．导致产品外观设计与产品的材料、工艺、结构、功能的矛盾急剧加深，解决两者之间的矛盾成为当务之急。包豪斯以它敏锐的视觉，针对性地提出了三个基本观点： 一、艺术与技术的统一； 二、设计的目的是人而不是产品； 三、设计要遵循自然和客观规律进行； 这些无疑体现出现代设计的观念和意识，具有鲜明的时代特征，也是对当时设计思潮的批判和否定。包豪斯构成理论是美学教育史上的一座丰碑。包豪斯构成理论教育的成功在于它的教育思想、教育审美产生了极大的凝聚力．吸引了许多在艺术上卓有建树的大师加盟，使包豪斯充满了活力和生气。在校长格罗佩斯旗下，先后有荷兰风格派代表人物杜斯

浏览器兼容性解决方案

所有浏览器通用 height: 100px; IE6 专用 _height: 100px; IE6 专用 *height: 100px; IE7 专用 *+height: 100px; IE7、FF 共用 height: 100px !important; 什么是浏览器兼容：当我们使用不同的浏览器（Firefox IE7 IE6）访问同一个网站，或者页面的时候，会出现一些不兼容的问题，在这种浏览器下显示正常，在另一种下就乱了，我们在编写CSS的时候会很恼火，刚修复了这个浏览器的问题，结果另外一个浏览器却出了新问题。好吧，我服了行吧，那我就利用你们的不兼容各写一段css,让他们各执行各的，呵呵。 一、!important (功能有限) 随着IE7对!important的支持, !important 方法现在只针对IE6的兼容.(注意写法.记得该声明位置需要提前.) 例如: #example { width: 100px !important; /* IE7+FF */ width: 200px; /* IE6 */ }

二、CSS HACK的方法(新手可以看看，高手就当路过吧) 首先需要知道的是： 所有浏览器通用 height: 100px; IE6 专用 _height: 100px; IE7 专用 *+height: 100px; IE6、IE7 共用 *height: 100px; IE7、FF 共用 height: 100px !important; 例如： #example { height:100px; } /* FF */ * html #example { height:200px; } /* IE6 */ *+html #example { height:300px; } /* IE7 */ 下面的这种方法比较简单 举几个例子： 1、IE6 - IE7+FF #example { height:100px; /* FF+IE7 */ _height:200px; /* IE6 */ } 其实这个用上面说的第一种方法也可以 #example { height:100px !important; /* FF+IE7 */ height:200px; /* IE6 */ } 2、IE6+IE7 - FF #example { height:100px; /* FF */ *height:200px; /* IE6+IE7 */ } 3、IE6+FF - IE7 #example { height:100px; /* IE6+FF */